OBJECTIVES/GOAL: To compare self-reported condom use problems and objectively determined semen exposure in 2 populations. STUDY DESIGN: Two randomized crossover trials in the United States and Brazil compared the failure rates of the female condom (FC) and male condom (MC). Participants used both condom types, completed condom-specific questionnaires to report problems, and collected precoital and postcoital samples of vaginal fluid. Prostate-specific antigen (PSA) was detected by immunoassay. RESULTS: Problems with condom use were reported less frequently in the Brazilian study (rate difference: FC = 24%, P <0.0001, MC = 5%, P = 0.003). By contrast, the PSA detection rates were similar for both the FC and the MC (rate difference: FC = 2%, MC = 1%, not significant). These results suggest that the PSA detection rate was similar in the 2 study groups and that self-reported problems may be a less reliable measure of condom failure. CONCLUSIONS: Use of biomarkers of condom failure like PSA may help to strengthen the validity of studies promoting behavior change for the prevention of sexually transmitted diseases. © Copyright 2007 American Sexually Transmitted Diseases Association.