Trials were conducted on a farm in Mpumalanga Province in South Africa to test the possibility of grading the colour of the ocular mucous membranes of sheep as an indication of the extent to which the animals are affected by Haemonchus contortus infection. The range of observed colour shades were classified into five categories, from red, through red-pink, pink and pink-white to white. Over a period of 125 days routine drenching of a flock of 388 sheep on irrigated kikuyu (Pennisetum clandestinum) pasture was terminated. During this time the animals were examined at practically weekly intervals and haematocrit determinations done for all the sheep with pale conjunctivae. Only those sheep having a haematocrit of 15% or lower were treated. Compared to a previous drenching tempo of close to every 3 weeks during the Haemonchus season on the farm, drenching was reduced by approximately 90%, as 70% of the sheep did not require salvage drenching and only 10% of the flock had to be given more than one salvage treatment. At the time of the trial the five clinical classifications were not related to predetermined haematocrit categories. However, when compared to categories that were set in later trials, 94% of the clinical estimates in the present trial were either in the correct haematocrit category, or, if not, the sheep were probably not disadvantaged by the errors. In 2.6% of cases the incorrect estimate may have placed the sheep concerned in jeopardy, as the haematocrit values were so low that salvage drenching was required, while the sheep were not regarded as anaemic. Changes in the mean haematocrit values of drenched and undrenched sheep were mirrored reciprocally by the changes in clinical colour estimates. Lactating ewes were by far the most susceptible class of sheep, as only 44.6% of them were able to manage without drenching, compared to 83% of dry, and 70.6% of pregnant ewes. Correlations between the haematocrits and clinical estimates were highly significant, although the associations were not high enough to give reasonable surety that the haematocrit values of individual animals could be predicted with confidence from their clinical classifications. Exceptionally large numbers of worms were recovered from seven of the 14 sheep that were culled because of age at the end of the trial, but these were reflected neither in their faecal worm egg counts, nor, with one exception only, in clinical signs.