To evaluate the utility of sperm tests as indicators of chemical effects on human spermatogenesis, the literature on 4 sperm tests used to assess chemically induced testicular dysfunction was reviewed. The tests surveyed included sperm count, motility, morphology (seminal cytology), and double Y-body (a fluorescence-based test thought to detect Y-chromosomal nondisjunction). There were 132 papers that provided sufficient data for evaluation. These reports encompassed 89 different chemical exposures: 53 were to single agents; 14 to complex mixtures; and 22 to combinations of 2 or more identified agents. Approximately 85% of the exposures were to experimental or therapeutic drugs, 10% were to occupational or environmental agents, and 5% were to drugs for personal use. The most common sperm parameter studied was sperm count (for 87 of the 89 exposures reviewed). Sperm motility was evaluated for 59 exposures, morphology for 44, and double Y-bodies for only 4. The 89 exposures reviewed were grouped into 4 classes: those which adversely effected spermatogenesis, as measured by one or more of the sperm tests (52); those suggestive of improving semen quality (11); those showing inconclusive evidence of adverse effects from exposure (14); and those showing no significant changes (12). Since the reviewed reports had a large variety of study designs, and since every attempt was made to include all reports with interpretable data, these classifications were based on reviewing committee decisions rather than on uniform statistical criteria. This review gives strong evidence that human sperm tests can be used to identify chemicals that affect sperm production, but because of our limited understanding of underlying mechanisms, the extent to which they can detect mutagens, carcinogens or agents that affect fertility remains uncertain. For the very few agents studied with both human and mouse sperm tests, similar test-responses were seen; thus sperm tests in mice and other laboratory mammals may have a potential role in hazard identification. An overall comparison of the 4 human sperm tests suggests that no one test is biologically more responsive than another; all of them may thus be needed when testing for chemically induced changes from agents of unknown activity. This review also gives evidence that sperm tests can be used to assess the extent and the potential reversibility of induced spermatogenic damage. The reviewing committee recommends further studies to determine (a) the dose-response characteristics of the human sperm tests, (b) details of the reversibility of induced changes with time after exposure, (c) the relative responses in the 4 sperm tests in exposed individuals, (d) the mechanism of action, (e) the biological and genetic implications of chemically induced effects, and (f) the comparison of responses among different species for risk assessment. The reviewing committee outlines specific considerations for planning new sperm studies on chemically exposed men.