Accurate measurement of semen exposure resulting from condom failures can refine public health messages and improve predictions of condom efficacy in preventing pregnancy and HIV transmission. Eight hundred and thirty couples enrolled in a condom efficacy study were asked to collect a baseline sample of ejaculate from the inside of the first study condom they used and to collect a postcoital vaginal sample whenever a study condom broke or slipped off during intercourse. All samples were quantitatively tested for prostate-specific antigen (PSA), a substance found only in human semen, using rocket immunoelectrophoresis, and inspected microscopically for presence of sperm. Sixty-eight baseline ejaculate samples collected from the inside of the first study condom by couples who subsequently experienced a condom failure averaged 13.4 microg PSA per swab and 79% of the samples averaged one or more sperm per high power field (hpf). Seventy-nine postcoital vaginal samples obtained after a condom break averaged 5.7 microg PSA per swab and only 38% averaged one or more sperm per hpf. The PSA results indicated a 50% reduction in semen exposure compared to baseline levels (p = 0.0001). Seventeen samples obtained after a condom slip-off averaged 2.5 microg PSA per swab and none of the samples averaged one or more sperm per hpf. The PSA results indicated an 80% reduction in semen exposure compared to baseline levels (p = 0.0001). Our results suggest that even condoms that fail reduce the risk of pregnancy and the transmission of sexually transmitted disease compared to unprotected intercourse. We also used PSA results to adjust a model designed to predict consistent-use pregnancy rates from condom breakage and slippage data.