We performed a prospective surveillance of semen quality among workers in the plant where 1,2-dibromo-3-chloropropane was first recognized as an occupational cause of impaired semen quality and of infertility. All male employees of the Agricultural Chemical Division were required to participate. Ninety-seven workers (92% participation) provided 258 semen samples over the 4 years of the program. Most samples were analyzed at the plant with a mini-laboratory designed for the study. Motility and shape measures were made objectively. Sixty-six subjects (68%) were non-azoospermic. Generalized multiple regression showed no significant predictors for any response, with the exception of the motility measures, which were reduced with longer times between ejaculation and assay. Between- and within-person standard deviations and correlations were calculated. Comparison of this population with fertile artificial insemination donors (16 men, 498 ejaculates) revealed generally higher ejaculate-to-ejaculate standard deviations in the worker samples. This is probably due to less well controlled conditions of sperm collection in the workplace setting. For cross-sectional studies, one ejaculate per worker is recommended as sufficient; for estimating an individual worker's mean, even three ejaculates may not provide enough precision.