This study represents the largest series to date documenting the gonadotoxic effect in humans of dibromochloropropane, a widely used pesticide. Three semen analyses, serum hormonal determinations (luteinizing hormone, follicle-stimulating hormone and testosterone) as well as genital examinations were completed for 228 workers at 2 chemical production sites and consisting of a dibromochloropropane-exposed and non-exposed cohort. Parameteric and non-parametric statistical analyses of the data sets of the sperm densities from the 2 subpopulations demonstrated statistical significance (p less than 0.10) at the short-term (1.5 years) manufacturing plant. Log transformation of the sperm count and hourly exposure data were necessary to develop meaningful statistical conclusions. The serum concentration of follicle-stimulating hormone as a group mean was significantly greater at both production sites for the exposed cohort when compared to the non-exposed participants but decreased by 10 levels of magnitude when the group demonstraing shorter but more recent exposure was compared to those from the plant with longer chemical production. Finally, a dose-response model suggested significant changes in sperm density at the short-term but more recently operated production site when more than 100 adjusted hours of exposure were exceeded, while the longer operated but longer closed facility demonstrated a significant impairment only when more than 1,000 adjusted hours of dibromochloropropane exposure were surpassed. This difference in exposure data may reflect regenerative changes in the tests once the gonadotoxic substance had been removed but exact nature of the dibromochloropropane effect and the possibility of a "no effect" concentration remain to be defined clearly.