First-catch early morning urine samples from patients attending a genitourinary medicine clinic were tested by three different enzyme immunoassays. The results suggest significant differences in the sensitivities and specificities of the different assays. The direct visualization of elementary bodies in urine deposits by direct immunofluorescence was used as the "gold standard," using a monoclonal antibody with a different epitope specificity from those of antibodies used in the enzyme-linked immunosorbent assays. We report for the first time that urine specimens represent an unbiased sample, free of the inherent sampling errors associated with other genital specimens. We feel that urine is a valid specimen for use in any evaluation study of new assays directed towards the detection of Chlamydia trachomatis.