Human semen contains not only spermatozoa but also other cells routinely differentiated as being peroxidase-positive (e.g. leucocytes) and peroxidase-negative (e.g. immature germ cells and lymphocytes) cells. Considerable uncertainty exists about their role in male fertility. To assess the clinical value of both parameters, and of microorganisms in semen, 391 treatments with in-vitro fertilization and embryo transfer were analysed retrospectively, and the concentrations of both peroxidase-positive and -negative cells, together with the presence of microorganisms in semen, were compared with both the fertilization and pregnancy rates. The data indicate that the results of treatment were affected only by excessively elevated concentrations of peroxidase-positive cells (> 6 x 10(6)/ml) and only marginally by the presence of microorganisms in the semen. The pregnancy rate after in-vitro fertilization and embryo transfer was not improved by antibiotic treatments preceding gamete recovery by several weeks. The increased presence of peroxidase-negative cells (e.g. germ cells) in semen was not associated with a significant change in the pregnancy rate. However, the concentration of peroxidase-negative cells in semen correlated significantly with sperm numbers (p < 0.01), sperm concentration (p < 0.01), and normal morphology rates (p < 0.01). It is concluded that short-term antibiotic treatment of asymptomatic patients before assisted reproduction should be handled with caution. The widespread view that peroxidase-negative cells in semen are harmful is rejected.