1. An investigation has been made into the accuracy of estimates of the proportions of dead spermatozoa in diluted and undiluted bull semen, using the nigrosineosin live-dead staining technique.
2. Two operators counted 24,673 spermatozoa, in lots of approximately 200, on smears derived from a number of subsamples taken from the same ejaculate.
3. The variations between different counts made on one smear were consistent with random sampling from a binomial population; the same applied to variations between slides made from one subsample; the variation between subsamples were, however, larger than would have been expected on the basis of random sampling.
4. Undiluted semen and semen diluted 1/4 with Ringer-phosphate-fructose solution had the same mean percentage dead and precision of estimation.
5. Tables are given showing how the precision of an estimate of the proportion of dead spermatozoa in a sample varies according to the percentage of dead spermatozoa, the total number of spermatozoa counted and according to the number of subsamples taken from the original ejaculate, each subsample being used for making counts.