Each of 3 technicians made duplicate pipettings from 46 semen specimens of 7 donors and filled 2 hemocytometer chambers from each pipette. The number of spermatozoa in each of the 4 large corner squares of the chamber were counted and recorded separately for a total of 2208 counts on 552 chambers. An analysis of variance was made and the percentage variances were calculated. Marked variation was found in counts among technicians and in duplicate counts by the same technician. The technician term and the related technician interaction terms accounted for 14.8 of the total study variance. Approximately one-half of the variance of the hemocytometer technique itself was due to the variance among technicians (57.4%) and one-half to the variance among duplicate hemocytometer determinations by the same technician (42.6%). The 95% confidence interval for hemocytometer counts, among technicians, on a single specimen which is equal to the mean of this study (28 million/ml), is [plus or minus]52%. The variance associated with the hemocytometer technique is large and must be considered in the design and analysis of studies of semen.