Infection with Chlamydia trachomatis is known to be a common cause of urethritis and cervicitis. The standard methods of detection require the collection of intra-urethral and/or cervical swabs, which may be submitted for culture, antigen detection or nucleic acid amplification. The collection of swabs is suitable only within the context of a health care facility. Recent reports have indicated that antigen detection can be used with urine specimens, and because these can be self-collected, this may be particularly useful for the detection of asymptomatic carriage. To determine the sensitivity and specificity of urine antigen assays in the detection of chlamydial infection in men. Two groups of men were investigated; men with urethritis attending clinics or private practitioners, and healthy adult men enrolled into either urban or rural HIV prevention projects. Urine samples from men in both groups were collected and assayed for the presence of chlamydial antigen using a commercial enzyme immunoassay (EIA) kit. For symptomatic men an intra-urethral swab was also collected and assayed for antigen detection using a commercial EIA. For asymptomatic men, a ligase chain reaction was carried out on the same urine sample. The prevalence of chlamydial antigen in symptomatic men was 15% (39/257), and in asymptomatic men was 4% (15/349). The sensitivity and specificity of urine EIA for symptomatic men was 87% and 83% respectively. For asymptomatic men, the sensitivity of urine EIA was 86%, and the specificity was 100%. Urine EIA is a relatively inexpensive method for the detection of chlamydial infections in men. The true specificity in symptomatic men may be higher, as the "gold standard" that we used may give false negative results. Antigen EIA for examination of urine specimens from community surveys of asymptomatic men may be particularly useful because of the low cost of assays, and because urine samples can be self-collected without discomfort to study subjects. The prevalence of C. trachomatis that we describe here is consistent with other studies of chlamydial epidemiology in Zimbabwe.