BRIEF REVIEW

International Journal of Sports Physiology and Performance, 2007;2:5-21

© 2007 Human Kinetics, Inc.

Understanding Sprint-Cycling
Performance: The Integration of Muscle
Power, Resistance, and Modeling
James C. Martin, Christopher J. Davidson,
and Eric R. Pardyjak

Sprint-cycling performance is paramount to competitive success in over half

the world-championship and Olympic races in the sport of cycling. This review
examines the current knowledge behind the interaction of propulsive and resistive
forces that determine sprint performance. Because of recent innovation in field
power-measuring devices, actual data from both elite track- and road-cycling
sprint performances provide additional insight into key performance determinants
and allow for the construction of complex models of sprint-cycling performance
suitable for forward integration. Modeling of various strategic scenarios using
a variety of field and laboratory data can highlight the relative value for certain
tactically driven choices during competition.

Key Words: exercise performance, physical performance, sport, sport physiology,

biomechanics, anaerobic

Of the 28 world-championship races governed by the Union Cycliste Interna-

tionale (UCI), 8 are all-out sprint events (menʼs and womenʼs sprint, 500/1000-m
time trial, Keirin, and BMX), 4 are often decided in the finishing sprint (menʼs and
womenʼs road race and scratch race), and 2 require repeated sprints (menʼs and
womenʼs points race). Thus, sprint performance is a major determinant of most
world-championship racing events. Ultimately, sprint performance represents equi-
librium of propulsive power and resistance. In this article, we will review factors
that influence maximal cycling power and cycling resistance and the limited inves-
tigations exploring sprint-bicycling performance. Sprint-performance articles are
quite limited because devices that accurately quantify power during actual bicycling
have only recently been developed and validated. Finally, we will integrate aspects
of muscle power and resistance in the context of sprint performance.

Aspects of Muscle Power

Maximal cycling power mainly depends on pedaling rate, muscle size, muscle-
fiber-type distribution, cycling position, and fatigue. Several investigators have

Martin and Davidson are with the Dept of Exercise and Sport Science, and Pardyjak, the Dept of
Mechanical Engineering, University of Utah, Salt Lake City, Utah 84112-0920.

03Martin(5).indd 5 2/23/07 3:07:04 PM

 6 Martin, Davidson, and Pardyjak

reported quadratic power–pedaling rate relationships for maximal cycling.1-5

The apex of that relationship is generally reported to occur at ~120 to 130 rev/
min1,2 (rpm), and power can vary by up to 25% within a range of pedaling rates
from 60 to 120 rpm, demonstrating the importance of this relationship (Figure
1). Furthermore, maximal power is highly impulsive, and instantaneous power
within each cycle can be up to 185% of power averaged over the entire cycle
(PREV). The highest values reported in the literature for power averaged over the
entire cycle are over 2500 W.6
Although pedaling rate is the most common velocity term for describing
cycling, it actually constrains 2 physiological phenomena. Pedaling rate, in conjunc-
tion with crank length, determines pedal speed and thereby sets shortening velocity
for uniarticular muscles that span the hip, knee, and ankle.7,8 Shortening velocity is
well known to influence muscle power9: Power will initially increase with increasing
shortening velocity, reach a maximum, and then decrease with further increases in
shortening velocity. In addition, pedaling rate, per se, sets the time within which
muscles must become excited, produce force while shortening, and relax before
lengthening. This time frame (half the time for a cycle, eg, 250 milliseconds at 120
rpm) reduces muscle force and power via excitation–relaxation kinetics.7,10,11 Are
both of these mechanisms, shortening velocity and excitation–relaxation kinetics,
important? Martin and colleagues7,12 have shown that the interaction of pedaling
rate and pedal speed constrains maximal cycling power across a broad range of
pedaling rates, pedal speeds, and crank lengths. Consequently, cycling power is
governed by the interaction of shortening velocity and excitation.7,13 A tangential

Figure 1 — Power–pedaling rate relationship. Data from a representative subject perform-

ing an inertial-load power test show instantaneous power (PI ⎯) and power averaged over
each complete revolution of the pedal cranks (PREV –ν–) in relation to pedaling rate (rpm).
The pedaling rate at which subjects reach a maximum value for PREV is defined as optimal
pedaling rate. Note that PI varies within each pedal revolution and reaches values up to
85% greater than PREV.

03Martin(5).indd 6 2/23/07 3:07:05 PM

 Sprint-Cycling Performance 7

finding in those studies was that cycle crank length was not a significant factor
during maximal efforts over the range of commercially available equipment (165
to 180 mm).12 Indeed, only large variations (50 mm) in crank length lead to small
but significant differences in the ability to produce maximal power.
Ultimately, cycling power must be produced by muscles that span the hip,
knee, and ankle. Power produced at each of those joints has not been reported in
peer-reviewed articles, but McDaniel et al14 have reported such data in an abstract.
Their data show that, during maximal seated isokinetic cycling at 120 rpm, 49% of
the power delivered to the pedal was produced at the knee, 32% was produced at
the hip, 9% was produced at the ankle, and 9% was transferred across the hip. This
warrants more research, and we are currently pursuing this in our laboratory.
Because most cycling power is produced at the knee and hip, it might reasonable
to expect that muscle mass about those joints influences maximal power. Recent
muscle-functional magnetic-resonance-imaging data showed that vastus lateralis
and vastus medialis volume accounted for 70% of the variance in measured maximal
power during repeated 6-second sprints.15 Similarly, lean thigh volume (r2 = .86)1
and lower limb volume (r = .92)16 have been reported to be highly predictive of
maximal power. Clearly, increased muscle mass will increase the power produced
by a muscle of any specific fiber-type distribution. In addition, fiber-type distribu-
tion dramatically alters maximal in situ muscle power. Muscles with predominantly
fast-twitch fibers exhibit higher maximum and optimal shortening velocities and
are up to 5 times as powerful (per unit muscle mass) as slow-twitch muscles.17
The relationship of optional velocity with fiber type is evident in cycling, as well.
Hautier and colleagues18 reported that optimal pedaling rate was highly correlated
(r = .88) with the proportion of cross-sectional area occupied by type II fibers
in vastus lateralis. Similarly, Pearson et al16 reported a high correlation (r = .80)
between the percentage of myosin heavy-chain II from a vastus lateralis biopsy
and optimal pedaling rate. Taking the combined effects of muscle mass and fiber-
type distribution a step further, Martin and colleagues19 reported that the product
of lean thigh volume and optimal pedaling rate (a surrogate marker for fiber-type
distribution) accounted for 83% of the variability in maximal power across the life
span (8 to 70 years), whereas lean thigh volume alone accounted for only 76%
of the variability. These data demonstrate that both muscle volume and fiber-type
distribution influence maximal cycling power.
During competitive sprinting, cyclists often adopt a standing position during
the initial (acceleration) phase and a seated position while at high speed.6 Reiser
et al20 reported that subjects produced greater power (~8%) when performing a
30-second Wingate anaerobic test in the standing position. Davidson et al21 have
observed (in an abstract) a slightly greater (12%) increase in maximal cycling power
during standing in a 3-second inertial-load test.1 If power is increased by standing,
where does that extra power come from? Davidson and colleagues22 reported (in an
abstract) that the increased power in the standing position resulted from additional
power from the upper body (transferred across the hip). It is interesting that ankle-,
knee-, and hip-joint powers did not differ with position in that investigation. These
data suggest that, although additional power might be generated in the standing
position, the biomechanics of power generation in the leg segments are unchanged.
This finding has implications for the positive influence of upper body strength and
dynamics during sprinting.

03Martin(5).indd 7 2/23/07 3:07:06 PM

 8 Martin, Davidson, and Pardyjak

Figure 2 — Schematic illustrating the flow around a cyclist and the surface stresses acting
on the rider that produce drag. The highest surface pressures are located at the leading edge
of the rider and the lowest pressures are in the wake. Form drag represents the integral of
this pressure over the entire surface area. Note that the rider is moving right to left and that
the coordinate system moves with the rider.

Thus far, this review has been focused on aspects of maximal power over short
durations, which are likely free from fatigue.23 Many competitive sprints take place
over 10 to 60 seconds, however, wherein power decreases because of fatigue. Fur-
thermore, fatigue can occur both within and between bouts of maximal sprinting
and depends on pedaling rate. Perhaps the most commonly reported protocol for
evaluating fatigue in a maximal cycling model is the Wingate anaerobic test.24 A
fatigue index is calculated from the maximum value recorded early in the test and
the minimum power produced at the end of the test. Normative data for Wingate
anaerobic tests indicate that fatigue index was approximately 38% and 35% in
young adult men and women,25 respectively, demonstrating the substantial effect
of fatigue. Interpretation of Wingate anaerobic test results is complicated because
pedaling rate varies throughout the test. That variability makes it difficult to know
how much of the “fatigue index” results from pedaling rate and how much is truly
representative of fatigue.
To control for the effects of pedaling rate, some authors have used isokinetic
cycling to study fatigue. Jones and colleagues26 reported that fatigue was greater
when pedaling at 140 rpm than at 60 rpm, but total work over 30 seconds did not
differ because power at 60 rpm was much lower early in the trial. Similarly, Beelen
and Sargeant23 reported that cyclists initially produced greater power at higher
pedaling rates but experienced greater fatigue at those higher rates. Data from
other protocols suggest that maximal efforts of longer duration (3 to 300 seconds)
might be influenced by a dependence on anaerobic metabolism.27-29 Weyand and
colleagues28,30 recently presented evidence for an essential metabolic effect on
fatigue progression and provided an extremely consistent exponential model of

03Martin(5).indd 8 2/23/07 3:07:06 PM

 Sprint-Cycling Performance 9

fatigue during maximal sprint cycling over times ranging up to 300 seconds. Their
model differs from traditional measures of maximal cycling, in that their EMG data
show progressive increases in recruitment over the duration of the trials, indicat-
ing that maximal effort might have only been achieved at termination of the trial.
Between bouts of short duration (<4 seconds) maximal sprinting power has been
shown to be highly repeatable and fatigue resistant31-33 with some dependence on
the recovery interval34,35 and activity.36 Subsequent maximal sprinting efforts might
actually be facilitated by a previous maximal effort33; thus the best “warm-up”
technique might be extremely task specific. To summarize, the ability to sustain
maximal cycling power is limited in duration, with large pedaling-rate-dependent
decrements in power beginning within the first few seconds, making fatigue an
important consideration for maximal sprint efforts longer than 4 seconds.
Finally, environmental conditions and substrate availability have been shown
to influence maximal cycling power. When sprints are repeated over time in stress-
ful environmental conditions, both increases in core and muscle temperature37-39
and declining hydration status37 reduce maximal power. In contrast, increasing
the availability of carbohydrate in blood37,40 and muscle41 attenuates the decline in
maximal power over time. Although significant differences are reported in these
studies, it is interesting to note that maximal sprint power was never reduced more
than 15% in any case. Indeed, the ability to maintain maximal power under stressful
conditions might be a genetic selection criterion in some species.42

Factors That Influence Resistance

in Sprint Cycling
Modeling Approaches
A number of cycling-performance models have been developed and are widely
used.43-46 They represent a simplified equation for the conservation of energy in rate
form; hence they are time-dependent balances between power production, power
sinks, and changes in energy states. Including time-dependent terms in modeling
sprint performance is critical because of the dynamic nature of sprinting. In this
section, we will use the model of Martin et al45 to exemplify the various aspects
of modeling sprint-cycling power. First, we will present the basic principles of
the model.
The basis for a cycling-performance model is Newtonʼs second law of motion
applied to a cyclist, namely,
(1)

where mT is the total mass of the rider and bicycle combined and and are the
acceleration and velocity vectors of the bicycle with respect to the ground. Many
forces act on the rider–bicycle combination, but the system is dominated by

(2)

Here, is the available force supplied to the pedals by the rider; , , and are
the aerodynamic, rolling-resistance, and bearing-resistance forces acting against

03Martin(5).indd 9 2/23/07 3:07:07 PM

 10 Martin, Davidson, and Pardyjak

the rider; and is the gravitational force. Equation 1 is a vector equation; hence,
it represents 3 equations in a Cartesian coordinate system. The 3 equations can be
collapsed into a single useful scalar mechanical-energy equation for the performance
of a cyclist by taking the dot product of Equation 1 with as follows:

(3)

Equation 3 is an expression for the time rate of change of the mechanical (kinetic)
energy associated with cycling ( ). The available power ( ) is some fraction of the
total energy rate expended by the athlete. It is related to the efficiency of the rider43
and represents a source term in Equation 3. The terms , , and are sinks
energy associated with aerodynamic, rolling, and bearing resistance. The last term
in Equation 3 ( ) is a potential energy term and can be either a source or a sink
depending on whether the rider is sprinting uphill or downhill. This term represents
the work rate associated with lifting the rider and bicycle mass.
Imbalances in the terms on the right-hand side of Equation 3 lead to increases
or decreases in rider velocity that are reflected in . For example, at the start of a
sprint, is greater than any of the resistance terms and results in rider acceleration
and increased kinetic energy. Later in the review, we discuss the ramifications of this
on sprint performance. In the following section, the details of each of the elements
of Equation 3 are reviewed, and simplified models are presented. Useful modeling
simplifications that are not covered here can be found elsewhere.43,44

Fundamentals of Aerodynamic Drag

When a cyclist is traveling over a flat surface at steady state, aerodynamic drag has
been shown to account for up to 96% of his or her available power.45 The aerodynamic
forces acting on a cyclist can be decomposed into 3 components associated with a Car-
tesian coordinate system fixed to the rider, as show in Figure 2. The net aerodynamic-
force component acting in the direction of travel is known as aerodynamic drag.
Formally, the drag force (FAero) is obtained by integrating the total surface stress ( )
over the surface area of the bicycle–cyclist (AS) combination, namely,
(4)

where is the unit vector opposite the direction of travel as shown in Figure 2. The
total surface stress is usually separated into the dominant surface stresses acting
parallel ( ) and normal (p) to the surface,

(5)

03Martin(5).indd 10 2/23/07 3:07:07 PM

 Sprint-Cycling Performance 11

Figure 3 — Mean power output and pedaling rate for men (n = 4, \BlTri\) and women (n
= 4, Δ) during a 200-m time trial. The data represent average power produced every second
by men and women and the corresponding pedaling rate. Within the time trial, cyclists
accelerated gradually for the first 5 to 7 seconds and then initiated maximum acceleration.
The parabolic curves represent the average predicted maximal power output (based on
the linear torque–pedaling rate regression) that could be produced in the range of pedal-
ing rates 75 to 170 rpm if the athletes were fatigue free. Data points below the parabolic
curve represent fatigue. The time trial was performed entirely on the descending limb of
the power–pedaling rate curve.

The first term in Equation 5 is usually referred to as skin friction, and the second
term is called form drag. For flow around objects such as cubes, cylinders, and
cyclists, pressure drag dominates the total drag.47 The flow around a cyclist is
mostly turbulent, with regions of separated flow and wakes that lead to increased
pressure drag.
For a control volume that wraps around the bicycle–cyclist combination, the
work rate associated with aerodynamic forces can be written as follows47:

(6)

where is the velocity of the wind with respect to the moving bicycle and is
the velocity of the bicycle.
Measuring the individual components of drag is very difficult, and the forces
are usually lumped together. Because of the large number of independent variables,
dimensional analysis is typically employed to reduce the number of experiments
that are needed to understand drag. If the atmosphere is assumed to be of constant
density for the small region where the bicycle is ridden, the drag force can be writ-
ten in the functional form FAero = f(ρ, μ, VW/B, L, e, Φ), where ρ is the air density
(a function of temperature, relative humidity, and barometric pressure48), μ is the
dynamic viscosity of air (primarily a function of temperature47), VW/B is the speed of

03Martin(5).indd 11 2/23/07 3:07:08 PM

 12 Martin, Davidson, and Pardyjak

the wind with respect to the bike, L is a length scale associated with the bike–rider
combination, e is a surface-roughness length scale, and Φ is the wind angle with
respect to the rider. Performing a dimensional analysis of the problem yields the
following dimensionless parameters that govern cyclist drag:

where A (~L2) is the projected frontal area of the rider. In the relationships given
above, CD is the drag coefficient and represents a nondimensional drag force, RL
is the Reynolds number and represents the ratio of inertial to viscous forces, and
k is a nondimensional roughness. The functional relationship simplifies to CD =
f(RL, k, Φ) and is different for any rider position and rider–bicycle combination (ie,
aerodynamic shape). This provides a framework for determining the drag force.
Most often CD is approximated to be independent of RL, k, and Φ. The total drag
force is then measured using a drag balance in a wind tunnel. If this is done, the
aerodynamic power term can be modeled as

Although many objects have CDs that do not vary much over a wide range
of wind speeds (spheres, cylinders, etc), the turbulent low-pressure region in the
wake of the body changes significantly with direction.45 Kyle49 and Zdravkovich et
al50 have measured CD using coast-down tests and wind-tunnel tests, respectively.
Kyle49 found a relatively constant CD of ~0.8 over a range of speeds (4.4 to 11.2
m/s) for riders in a racing position (hands on the dropped part of the handle bar,
arms stiff, and head up) and slight reductions (CD ≈ 1.09 to 0.92) with increasing
speed for a rider standing vertically on the pedals. Zdravkovich et al50 reported
smaller CDs than Kyle49 but also found very little variation in CD for large Reyn-
olds numbers. As expected, both studies found that CD varied significantly with
position and size of riders. Olds et al48 also found lower CD values than Kyle for a
track cyclist in the drops by approximately a factor of 2. The CD values reported
by Kyle,49 Zdravkovich et al,50 and Olds et al48 range from ~0.35 to 1.1, with the
lowest values for riders in tucked descending and time-trial positions and the largest
values for standing on the pedals.
Calculating CD requires estimates of rider frontal area using photographic
weighing,43 digital photo analysis,51,52 or morphometric parameterization.46 Great
care must be taken to minimize uncertainty in area measurement. To eliminate this
difficulty, Martin et al45 recommend using CDA, or “drag area” (reported in m2).
Drag area effectively integrates all the effects associated with stresses acting on
the rider and is dominated by the turbulence associated with rider position, shape,
size, and surface roughness. CDA can be measured in a wind tunnel45 or on the road
using commercial power meters,6 tow tests,43 and coast-down tests.49 Martin et al6
found that wind-tunnel and field-derived CDAs using power meters did not differ.
This finding allowed them to use this field technique to derive CDAs for sprinters
in both seated and standing positions.
Drag reduction is of utmost importance to sprint cyclists because hundredths
of a second are important in deciding elite-level competitions. Typically, reductions

03Martin(5).indd 12 2/23/07 3:07:09 PM

 Sprint-Cycling Performance 13

are realized by optimizing rider position and equipment if riding alone and position-
ing (ie, drafting) when riding in groups. Independent drag data on equipment have
been presented by Kyle,53 Greenwell et al,54 Tew and Sayers,55 and Martin et al.6,45
The aerodynamic benefit of riding in a group of cyclists is more poorly understood.
Kyle49 reported an ~33% reduction in the power required to maintain a specified
speed and an ~38% reduction in wind resistance for riders in a 2-man pace line.
These experiments included coast-down tests over a wide range of speeds (24 to 56
km/h) and wheel-to-wheel distances (Δx ≈ 0 to 2 m). Kyle also found no additional
reductions for drafting behind additional riders. Zdravkovich50 reported similar
wind-resistance reductions of 37% and 35% for a 2-man pace line with wheel-
to-wheel distances of 70 and 90 cm, respectively, at 30 km/h. Using field-derived
power measurements for a 4000-m team pursuit at 60 km/h, Broker et al56 reported
an ~29% decrease in the required power for the second rider in the pursuit. They
also reported an additional reduction of 7% in required power for the third rider,
with no measurable reductions for the fourth rider. Both Kyle49 and Broker et al56
allude to the importance of a skill component in drafting to obtain the maximum
benefits and that the reductions can vary significantly for less skilled riders.
Complete models regarding drafting are still lacking. Olds et al48 have deduced
the following aerodynamic-resistance correction factor caused by drafting by piec-
ing together data sets:

CFdraft = 1 – 0.3835 + 1.25Δx + 0.0405Δx2

Unfortunately, a model that modifies a riderʼs measured CDA as a function

of wheel separation, effective wind angle, and sideways distance in a pace line
remains illusive.

Rolling and Bearing Resistance

The force caused by rolling resistance is related to the weight of the bike and rider,
wheel radius, tire pressure, tread pattern, casing construction, and the gradient and
texture of the riding surface. Following Olds et al,44 the rolling resistance is modeled
as the normal force times a coefficient of rolling resistance (CRR) that includes the
effects of tire and surface characteristics. This simplification leads to a model of
the form , where θ is the local angle of the road surface given by
the arctangent of the grade of the road and g is the acceleration of gravity. Kyle53
reports rolling resistances for different road and track tires and surfaces ranging
from 0.0019 to 0.0040.
The bearing resistance is typically extremely small (2% to 5% of total power)
and is a function of bearing type, load, and angular velocity of the wheel.53 Using
the work of Dahn and colleagues,57 Martin et al45 estimated the bearing friction for
a typical cyclist with cartridge bearings to be

This method is somewhat limited by the ability to find reported data on the exact
bearing type used in a bicycle. To address that limitation, Martin et al6 recently
combined both rolling and bearing resistance into a single global coefficient of
resistance that was determined in the field using multiple linear regression.

03Martin(5).indd 13 2/23/07 3:07:10 PM

 14 Martin, Davidson, and Pardyjak

A Field Model
Combining these simplifications leads to the following useful field model that can
be used for steady-state45 or sprint cycling6:

Here, Ec is a chain drive efficiency term. As shown in the preceding equation,

Martin et al45 add 2 terms related to wheel rotation: an aerodynamic force associ-
ated with wheel rotation (Fw) and a wheel-rotation kinetic-energy term I/r2, where
I is moment of inertia of the wheel.
Martin et al6 have successfully modeled sprint cycling on the track using this
methodology. They point out the importance of modeling the corners in the track
and including the difference between speed of center of mass of the rider and the
wheels because of lean angle. For sprint cycling, additional work remains to be
done to further our understanding of the complex changes in the drag area while
going from a seated to standing position and in describing drag reductions caused
by drafting in a variety of scenarios.

Sprint Performance
In this final section, studies of sprint-bicycling (as opposed to laboratory-based cycle
ergometry) performance and interaction of factors that influence power production
and resistance are examined.
Only a few authors have reported measures recorded during actual sprint
bicycling. Craig and colleagues27 measured cycling power during ergometer tests
of 10 to 60 seconds and correlated those measures with 1000-m bicycling-perfor-
mance time. The only significant predictors of 1000-m performance time were
power and work performed in the 60-second ergometer test. They suggested that
those results were caused by the increased reliance on aerobic metabolism during
the 60-second tests and the 1000-m performance.29,30 That is, performances of
durations greater than 3 seconds are determined by both maximum muscle power
and aerobic power.30
In a similar investigation, Dorel and colleagues51 reported laboratory and
competition data of world-class sprint cyclists. Maximum power, normalized by
frontal surface area, was significantly correlated with 200-m performance velocity,
suggesting that performance largely depended on the ridersʼ ability to overcome
aerodynamic drag. Furthermore, average pedaling rates during the 200-m (155 ±
3 rpm) were significantly greater than pedaling rates for maximum power (130
± 5 rpm). The authors speculated that these cyclists chose smaller gear ratios in
competition to optimize power during acceleration. In support of that speculation,
Martin and colleagues58 reported in an abstract that world-class sprint cyclists
performed the entire 200-m time trial on the descending limb of the power–
pedaling rate relationship (Figure 3). They reached pedaling rates (163 ± 3 rpm) that

03Martin(5).indd 14 2/23/07 3:07:10 PM

 Sprint-Cycling Performance 15

would reduce power by approximately 35% at maximum speed. Finally, Gardner

and colleagues59 recorded power during national and international competition and
reported that elite sprint cyclists produced peak power values of 1939 ± 241 W
during match sprint competitions. Peak power was produced at pedaling rates of
130 ± 10 rpm, and the cyclists reached peak pedaling rates of 161 ± 3 rpm. Thus,
both the aerodynamic and acceleration terms are critical for 200-m bicycling
performance, and selected gear ratios might represent a compromise between the
need for power during acceleration and at maximum speed. In addition, the aerobic
contribution to sprint performances can be significant. Although frontal area is an
important predictor of average velocity, aerodynamic-drag area should be even more
predictive. Instantaneous measures of power and speed would provide additional
insight into performance power.
Recently, Martin and colleagues6 reported that aerodynamic-drag area and
rolling resistance could be accurately determined from field bicycling trials with
power measurement. Those field-based parameters were then used to model sprint-
bicycling speed by forward integration. Although the focus of that article was the
modeling technique, the data represented maximal performances from a standing
start by world- and Olympic-champion athletes. Furthermore, this model allows
calculation of the power associated with acceleration and aerodynamic drag through-
out each performance and can be used to determine the relative importance of those
resistance terms. Qualitatively, more power is likely associated with acceleration at
the start, whereas at high speed more power is associated with aerodynamic drag.

Figure 4 — Power, resistance, and energy changes during a 1000-m time-trial performance
of a world-champion cyclist. Speed (\BlTri\) represents the speed of the center of mass,
which does not vary in the turns as does wheel speed. In the first few seconds, power (Δ)
increased and then decreased, reflecting the power–pedaling rate relationship. Changes in
kinetic energy (❑) required most of the generated power for the first 12 seconds, whereas
aerodynamic resistance (ν) dominated thereafter. The discontinuity in aerodynamic-drag
power represents the standing-to-seated transition. After 21 seconds, the rider slowed and
energy was negative, meaning that stored kinetic energy was being given up to do external work.
Rolling resistance power (——) varied because of increased normal force in the turns.

03Martin(5).indd 15 2/23/07 3:07:10 PM

 16 Martin, Davidson, and Pardyjak

To our knowledge, however, no authors have addressed this issue quantitatively.

It might be useful to reexamine these data to determine the power required for
acceleration and aerodynamic drag during a 1000-m time trial (Figure 4). Within
the first 12 seconds, power increased to a maximum and then decreased, reflecting
the power–pedaling rate relationship and possibly some early fatigue. Acceleration
required 73% of the work over the first 12 seconds. By 12 seconds this 1000-m
time-trial cyclist had exceeded 17 m/s, and aerodynamic drag dominated the power
requirement from then until the end of the performance, accounting for 74% of
total work. Twelve seconds was also when the cyclist transitioned from standing to
seated cycling. Speed increased until 17 seconds even though power had decreased
substantially. From 17 to 21 seconds, speed was nearly constant, indicating that
power production and resistance were matched. Beyond 21 seconds, the cyclist
began to slow, and thus was negative, indicating that kinetic energy was being
given up to do external work. Consequently, accounted for only 17% of work
for the entire performance. Rolling resistance accounted for 9% of total work,
which was surprising given the low CRR (0.24%). This relatively high contribution
resulted partially from the increased normal force in the curves. The standing-start
nature of the 1000-m time trial dictates that substantial work must be associated
with acceleration. Other sprinting performances, such as road-race finishes, are
initiated from high speed and likely require less acceleration work.
To date, no authors have reported power data for a finishing sprint in a profes-
sional road race; however, we recently obtained data from the winner of a road stage
in a major national tour and UCI “pro tour” event (Figure 5). The final maximal
effort was 14 seconds in duration, during which average power was 926 W and
peak power was 1097 W. In the final sprint the cyclist accelerated from 58.2 to 65
km/h (16.2 to 18.1 m/s), indicating an increase in kinetic energy of 2586 J (total
mass was 80 kg). During this period, total work was 12,960 J, and, consequently,
inertial power represented 20% of total work output (184 W for 14 seconds).
Assuming CRR = 0.4%, rolling resistance required 54 W, and 688 W (74% of power
during the finishing sprint) remained to overcome aerodynamic resistance. This
power indicates a drag area of approximately 0.288 m2, suggesting that the rider
was in a good aerodynamic position considering he used traditional handlebars.60
Thus, in this UCI pro-tour wining sprint, the riderʼs power was distributed as 74%
for aerodynamic resistance, 20% for , and 6% for rolling resistance. As a final
comment, the peak power of 1097 W might seem modest compared with the power
reported for track cyclists over short distances. This power, however, won a major
professional road stage race with a large peloton intact. It was produced after 5 and
a half hours of racing and was the culmination of several kilometers of increasing
speed in approaching the finish line. During the final 3 minutes, this cyclist averaged
490 W and exceeded 600 W for 64 seconds. Thus, this performance represents an
extremely high standard for professional road-sprinting power.
Other factors interactively influence power production and resistance. Stand-
ing to pedal increases power output 8% to 12%20,21 and aerodynamic-drag area
by 24%.6 What if the road cyclist described had stood up to sprint? Similarly,
increased muscle mass directly increases power output and inertia and indirectly
increases aerodynamic-drag area. Is a bigger sprinter likely to win against a smaller
sprinter with the same power-to-mass ratio? To model these scenarios, an idealized
power–time data set, based on the data obtained from the road-sprint cyclist just

03Martin(5).indd 16 2/23/07 3:07:11 PM

 Sprint-Cycling Performance 17

Figure 5 — Power and speed during the final 3 minutes of a pro-tour road event. Data are
from the stage winner. Power (❑) for the last 3 minutes averaged 490 W and was greater
than 600 W for 64 seconds. The final sprint lasted 14 seconds, during which average power
was 926 W, peak power was 1097 W, and speed (■) increased from 16.2 to 18.1 m/s (58.2
to 65 km/h). The variability of power in the final minute attests to the frenzied nature of
the final sprint lead-out.

described, was used. The estimated aerodynamic-drag area of that cyclist (0.288
m2) and coefficient of rolling resistance (0.4%), mass (80 kg), and starting speed
(58.2 km/h) were used. The forward integration model was used to predict speed
and distance for 14 seconds. The standing cyclist was assumed to produce 10%
more power, but his drag area increased to 0.36 m2. The modeled “increased mass”
cyclist was 10% more massive and powerful, and his drag area was increased by
7.4% (1.10.75). The model predicted that the standing cyclist would have lost a
remarkable 4.1 m compared with his seated performance because the increase in
power did not compensate for the increased aerodynamic drag. In contrast, the larger
rider would have defeated his smaller counterpart by 95 cm because the increase in
power was slightly greater than the increase in aerodynamic drag. Another factor that
interactively influences power and resistance during sprint cycling is altitude. Alti-
tude (air density) will decrease aerodynamic resistance, but the influence of altitude
on fatigue in sprinting performance is not well known. Consequently, the effects of
altitude will not be modeled here but remain an important area for future research.
Cyclists often take the rear position at the start of a sprint. This strategy
decreases aerodynamic drag but has the disadvantage that the following cyclist
must pass the leading rider to win. To model that strategy it was assumed that 2
cyclists produced the same power but one rider started 1 m behind the otherʼs rear
wheel (~2.8 m behind the lead riderʼs front wheel). The drafting riderʼs drag area
was conservatively assumed to be reduced by 25%49 until he drew even with the
lead cyclistʼs rear wheel and moved out of the draft. By initiating acceleration in
the draft, the following rider increased speed much more rapidly, and he passed
the lead rider within 8 seconds (Figure 6) and won by 1.05 m (over 236 m). In this

03Martin(5).indd 17 2/23/07 3:07:11 PM

 18 Martin, Davidson, and Pardyjak

Figure 6 — The effects of drafting on sprint performance. (a) The following rider accel-
erates (■) at a greater rate while drafting the lead (❑) rider. (b) The greater speed attained
while drafting allows him to pass and win by over 1 m.

example a conservative but fixed reduction in aerodynamic-drag area during drafting

was assumed. A more sophisticated approach could include terms for the baseline
aerodynamic-drag area of both riders, the gap between the 2, and any sideways
offset. This type of interactive competitive approach might be the most exciting
area for advancements in performance modeling and could identify strategies for
novice and world-class competitors.

Summary
Maximal cycling power is influenced by pedaling rate, muscle size and fiber
composition, and fatigue. Cycling speed is resisted by aerodynamic and rolling

03Martin(5).indd 18 2/23/07 3:07:12 PM

 Sprint-Cycling Performance 19

friction, and any imbalance in applied versus required power results in changes
in system energy. Sprinting performances arise from interaction of power produc-
tion, resistance, and changes in energy. The relative contribution of the resistance
and energy varies, with energy changes or acceleration dominating at low speed
and aerodynamics dominating at high speed. Areas ripe for future study include
improved models of drag reduction from drafting, optimization of power or pedal-
ing rate for short time-trial efforts, interactive effects of fatigue and pedaling rate
for sprints longer than 4 seconds, and maximal power reductions associated with
high-altitude performance.

Acknowledgments
The authors wish to express their sincere thanks to Stefan Vogt and Olaf Schumacher for
their valuable help in the preparation of this review.

References
1. Martin JC, Wagner BM, Coyle EF. Inertial-load method determines maximal cycling
power in a single exercise bout. Med Sci Sports Exerc. 1997;29:1505-1512.
2. Sargeant AJ, Hoinville E, Young A. Maximum leg force and power output during
short-term dynamic exercise. J Appl Physiol. 1981;51(5):1175-1182.
3. McCartney N, Heigenhauser GJ, Jones NL. Power output and fatigue of human muscle
in maximal cycling exercise. J Appl Physiol. 1983;55(1):218-224.
4. Buttelli O, Seck D, Vandewalle H, Jouanin JC, Monod H. Effect of fatigue on maximal
velocity and maximal torque during short exhausting cycling. Eur J Appl Physiol.
1996;73(1-2):175-179.
5. Seck D, Vandewalle H, Decrops N, Monod H. Maximal power and torque–velocity
relationship on a cycle ergometer during the acceleration phase of a single all-out
exercise. Eur J Appl Physiol. 1995;70(2):161-168.
6. Martin JC, Gardner AS, Barras M, Martin DT. Modeling sprint cycling using field-derived
parameters and forward integration. Med Sci Sports Exerc. 2006;38(3):592-597.
7. Martin JC, Brown NA, Anderson FC, Spirduso WW. A governing relationship for
repetitive muscular contraction. J Biomech. 2000;33(8):969-974.
8. Yoshihuku Y, Herzog W. Optimal design parameters of the bicycle-rider system for
maximal muscle power output. J Biomech. 1990;23(10):1069-1079.
9. Hill AV. The heat of shortening and dynamic constants of muscle. Proc R Soc Lond B
Biol Sci. 1938;126:136-195.
10. van Soest O, Casius LJ. Which factors determine the optimal pedaling rate in sprint
cycling? Med Sci Sports Exerc. 2000;32(11):1927-1934.
11. Caiozzo VJ, Baldwin KM. Determinants of work produced by skeletal muscle: poten-
tial limitations of activation and relaxation. Am J Physiol. 1997;273(3 pt 1):C1049-
C1056.
12. Martin JC, Spirduso WW. Determinants of maximal cycling power: crank length,
pedalling rate, and pedal speed. Eur J Appl Physiol. 2001;84(5):413-418.
13. Martin JC. Muscle power: the interaction of cycle frequency and shortening velocity.
Exerc Sport Sci Rev. In press.
14. McDaniel J, Tomas A, Hunter EL, et al. Joint power distribution at 60, 90, and 120 rpm
during seated maximal cycling abstract. Med Sci Sports Exerc. 2005;37(S5):S123.
15. Akima H, Kinugasa R, Kuno S. Recruitment of the thigh muscles during sprint cycling
by muscle functional magnetic resonance imaging. Int J Sports Med. 2005;26(4):245-
252.

03Martin(5).indd 19 2/23/07 3:07:12 PM

 20 Martin, Davidson, and Pardyjak

16. Pearson SJ, Cobbold M, Orrell RW, Harridge SD. Power output and muscle
myosin heavy chain composition in young and elderly men. Med Sci Sports Exerc.
2006;38(9):1601-1607.
17. Swoap SJ, Caiozzo VJ, Baldwin KM. Optimal shortening velocities for in situ power
production of rat soleus and plantaris muscles. Am J Physiol. 1997;273(3 pt 1):C1057-
C1063.
18. Hautier CA, Linossier MT, Belli A, Lacour JR, Arsac LM. Optimal velocity for maximal
power production in non-isokinetic cycling is related to muscle fibre type composition.
Eur J Appl Physiol. 1996;74:114-118.
19. Martin JC, Farrar RP, Wagner BM, Spirduso WW. Maximal power across the lifespan.
J Gerontol A Biol Sci Med Sci. 2000;55(6):M311-M316.
20. Reiser RF, Maines JM, Eisenmann JC, Wilkinson JG. Standing and seated Wingate
protocols in human cycling: a comparison of standard parameters. Eur J Appl Physiol.
2002;88(1-2):152-157.
21. Davidson CJ, Wagner BM, Martin JC. Seated and standing maximal neuromuscular
cycling power [abstract]. Med Sci Sports Exerc. 2004;36(5S):S344.
22. Davidson CJ, Horscroft RD, McDaniel J, et al. The biomechanics of producing stand-
ing and seated maximal cycling power [abstract]. Med Sci Sports Exerc. 2005;37(S5):
S393.
23. Beelen A, Sargeant AJ. Effect of fatigue on maximal power output at different contrac-
tion velocities in humans. J Appl Physiol. 1991;71(6):2332-2337.
24. Bar-Or O, Dotan R, Inbar O. A 30-second all-out ergometric test: its reliability and
validity for anaerobic capacity. Isr J Med Sci. 1977;13:126.
25. Maud PJ, Shultz BB. Norms for the Wingate anaerobic test with comparison to another
similar test. Res Q Exerc Sport. 1989;60(2):144-151.
26. Jones NL, McCartney N, Graham T, et al. Muscle performance and metabolism in maximal
isokinetic cycling at slow and fast speeds. J Appl Physiol. 1985;59(1):132-136.
27. Craig NP, Pyke FS, Norton KI. Specificity of test duration when assessing the anaerobic
lactacid capacity of high-performance track cyclists. Int J Sports Med. 1989;10(4):237-242.
28. Bundle MW, Ernst CL, Bellizzi MJ, Wright S, Weyand PG. A metabolic basis for
impaired muscle force production and neuromuscular compensation during sprint
cycling. Am J Physiol Regul Integr Comp Physiol. 2006;291(5):R1457-R1464.
29. Medbo JI, Tabata I. Relative importance of aerobic and anaerobic energy release during
short-lasting exhausting bicycle exercise. J Appl Physiol. 1989;67(5):1881-1886.
30. Weyand PG, Lin JE, Bundle MW. Sprint performance–duration relationships are set by
the fractional duration of external force application. Am J Physiol Regul Integr Comp
Physiol. 2006;290(3):R758-R765.
31. Watt KK, Hopkins WG, Snow RJ. Reliability of performance in repeated sprint cycling
tests. J Sci Med Sport. 2002;5(4):354-361.
32. Mendez-Villanueva A, Bishop D, Hamer P. Reproducibility of a 6-s maximal cycling
sprint test. J Sci Med Sport. 2006 Aug 31.
33. Martin JC, Diedrich D, Coyle EF. Time course of learning to produce maximum cycling
power. Int J Sports Med. 2000;21(7):485-487.
34. Glaister M, Stone MH, Stewart AM, Hughes M, Moir GL. The influence of recovery dura-
tion on multiple sprint cycling performance. J Strength Cond Res. 2005;19(4):831-837.
35. Buttelli O, Vandewalle H, Jouanin JC. Recovery of the torque–velocity relationship after
short exhausting cycling exercise. Eur J Appl Physiol Occup Physiol. 1999;80(3):249-251.
36. Bogdanis GC, Nevill ME, Lakomy HK, Graham CM, Louis G. Effects of active recovery
on power output during repeated maximal sprint cycling. Eur J Appl Physiol Occup
Physiol. 1996;74(5):461-469.
37. Fritzsche RG, Switzer TW, Hodgkinson BJ, Lee SH, Martin JC, Coyle EF. Water and
carbohydrate ingestion during prolonged exercise increase maximal neuromuscular
power. J Appl Physiol. 2000;88(2):730-737.

03Martin(5).indd 20 2/23/07 3:07:13 PM

 Sprint-Cycling Performance 21

38. Drust B, Rasmussen P, Mohr M, Nielsen B, Nybo L. Elevations in core and muscle
temperature impairs repeated sprint performance. Acta Physiol Scand. 2005;183(2):181-
190.
39. Castle PC, Macdonald AL, Philp A, Webborn A, Watt PW, Maxwell NS. Precooling leg
muscle improves intermittent sprint exercise performance in hot, humid conditions. J
Appl Physiol. 2006;100(4):1377-1384.
40. Ball TC, Headley SA, Vanderburgh PM, Smith JC. Periodic carbohydrate replacement
during 50 min of high-intensity cycling improves subsequent sprint performance. Int
J Sport Nutr. 1995;5(2):151-158.
41. Rockwell MS, Rankin JW, Dixon H. Effects of muscle glycogen on performance of re-
peated sprints and mechanisms of fatigue. Int J Sport Nutr Exerc Metab. 2003;13(1):1-14.
42. Husak JF, Fox SF, Lovern MB, Van Den Bussche RA. Faster lizards sire more off-
spring: sexual selection on whole-animal performance. Evolution Int J Org Evolution.
2006;60(10):2122-2130.
43. di Prampero PE, Cortili G, Mognoni P, Saibene F. Equation of motion of a cyclist. J
Appl Physiol. 1979;47(1):201-206.
44. Olds TS, Norton KI, Craig NP. Mathematical model of cycling performance. J Appl
Physiol. 1993;75(2):730-737.
45. Martin JC, Milliken DL, Cobb JE, McFadden KL, Coggan AR. Validation of a math-
ematical model for road cycling power. J Appl Biomech. 1998;14(3):276-291.
46. Bassett DR Jr, Kyle CR, Passfield L, Broker JP, Burke ER. Comparing cycling world
hour records, 1967–1996: modeling with empirical data. Med Sci Sports Exerc.
1999;31(11):1665-1676.
47. Potter MC, Foss JF. Fluid Mechanics. Okemos, Mich: Great Lakes Press; 1982.
48. Olds TS, Norton KI, Lowe EL, Olive S, Reay F, Ly S. Modeling road-cycling perfor-
mance. J Appl Physiol. 1995;78(4):1596-1611.
49. Kyle CR. Reduction of wind resistance and power output of racing cyclists and runners
traveling in groups. Ergonomics. 1979;22(4):387.
50. Zdravkovich MM, Ashcroft MW, Chisholm SJ, Hicks N. Effect of cyclistsʼ posture and
vicinity of another cyclists on aerodynamic drag. In: Haake S, ed. The Engineering of
Sport. Rotterdam, The Netherlands: Swets & Zeitlinger; 1996:356.
51. Dorel S, Hautier CA, Rambaud O, et al. Torque and power–velocity relationships in
cycling: relevance to track sprint performance in world-class cyclists. Int J Sports Med.
2005;26(9):739-746.
52. Heil DP. Body mass scaling of frontal area in competitive cyclists not using aero-
handlebars. Eur J Appl Physiol. 2002;87(6):520-528.
53. Kyle CR. Energy and aerodynamics in bicycling. Clin Sports Med. 1994;13(1):39-
73.
54. Greenwell DI, Wood NJ, Bridge EKL, Addy RJ. Aerodynamic characteristics of low-
drag bicycle wheels. Aeronaut J. 1995;99(983):109-120.
55. Tew GS, Sayers AT. Aerodynamics of yawed racing cycle wheels. Wind Eng Ind Aero-
dynamics. 1999;82(1-3):209-222.
56. Broker JP, Kyle CR, Burke ER. Racing cyclist power requirements in the 4000-m
individual and team pursuits. Med Sci Sports Exerc. 1999;31(11):1677-1685.
57. Dahn K, Mai L, Poland J, Jenkins C. Frictional resistance in bicycle wheel bearings.
Cycling Sci. 1991(Dec and Sept):28-32.
58. Martin JC, Gardner AS, Barras M, Martin DT. Power, pedaling rate, and fatigue during
the 200 meter time trail performance [abstract]. Med Sci Sports Exerc. 2005;37(5S):
S85-S86.
59. Gardner AS, Martin DT, Barras M, Jenkins DG, Hahn AG. Power output demands of
elite track sprint cycling. Int J Perf Anal Sport. 2005;5:149-154.
60. Martin JC, Cobb JE, Jeukendrup AE. Body position and aerodynamics. In: Jeukendrup
AE, ed. High Performance Cycling. Champaign, Ill: Human Kinetics; 2002:103-112.

03Martin(5).indd 21 2/23/07 3:07:13 PM