In freestyle swimming the arm action is routinely quantified by stroke count and rate, yet no method is currently available for quantifying kick. In this study, we assessed the validity and reliability of inertial sensor technology (gyroscope) to assess kick count and rate. Twelve Paralympic swimmers completed a 100-m freestyle-swimming time-trial and freestyle kicking-only time-trial three times each in a season. An algorithm was developed to detect the up and down beat of individual kicks from the gyroscope trace. For comparative purposes, underwater video analysis provided the criterion measure. The standard error of the estimate (validity) for kick count, expressed as a coefficient of variation, was 5.9% (90% confidence interval 5.5 to 6.4) for swimming, and 0.6% (0.5 to 0.6) for kicking-only trials. The mean bias for kick count was -1.7% (-2.4 to -1.1) for swimming, and -0.1% (-0.2 to -0.1) for kicking-only trials. Correlations between the sensor and video for kick count were 0.96 (0.95 to 0.97) for swimming, and 1.00 (1.00 to 1.00) for kicking-only trials. The typical error of the measurement (reliability) between trials was approximately 4% for kick count and rate. The inertial sensors and associated software used generated sufficient validity and reliability estimates to quantify moderate to large changes in kick count and rate in freestyle swimming.