Abstract: The aim of this study was to increase the precision of assessment of the number of round cells observed in the peroxidase test for detection of seminal leukocytes（granulocytes）. The dilution of semen was reduced and the volume of suspension examined was increased for semen samples containing between 0.6 and 6 million round cells per mL. A1+5（1∶6）dilution was compatible with measurable peroxidase activity and a sufficiently clear background for cell assessment. At this dilution，and with assessment of all 18 grids on both sides of the Neubauer-improved counting chamber，only three of the 10 samples （nominal cell concentrations of 1.9×10(6)-3.3×10(6) mL-1) presented 400 round cells or more. As lower seminal dilutions were incompatible with easy detection of round cells or their peroxidase reaction product，it was not possible to provide precise measurements（sampling error 5%）of the suggested lower reference limit of 1伊106 cells per mL. The results indicate that this poor precision of measuring 1×10(6) round cells per mL could explain the discrepant reports on the acceptability of the cut-off values for leukocytospermia. Such reference limits need to be established with statistically sound methods.

Key words: Cell number, Peroxidase, Precision, Round cells, Semen