Seminal WBC counts obtained by an mAb-based immunohistologic method correlated well with seminal granulocyte counts obtained with a simple peroxidase method (rho = +0.70; P < 0.0001). However, total WBC counts were significantly higher than granulocyte counts for most samples. With the immunohistologic method, 17 of 112 samples (15.2%) contained > 10(6) WBC/mL semen, whereas the peroxidase method resulted in only 10 samples (8.9%) with > 10(6) WBC/mL. When the threshold defining leukocytospermia was set at 1 x 10(6) positive cells/mL for both methods, the specificity of the peroxidase test compared with the immunohistology technique was 100% (10/10), but the sensitivity was only 58.8% (10/17). When the threshold for leukocytospermia in the peroxidase test was lowered to 5 x 10(5) positive cells/mL semen, the sensitivity relative to the immunohistology technique increased to 94.1% (16/17), and specificity remained 100% (16/16). Likewise, good interassay sensitivity and specificity values were obtained with thresholds of 10(6) WBC/mL for the peroxidase assay and 2 x 10(6) WBC/mL for the immunohistology assay. We conclude that either peroxidase or immunohistology assays can be used to screen for leukocytospermia, but that more research is needed to establish thresholds for pathological levels of WBC in semen using these two approaches. Total round cell counts are of no value for enumerating WBC in semen.