AIDS-related research relies primarily on self-reports of sexual practices. Therefore, determining which data collection methods yield more candid information is critical. Data from a study of gay men's sexual adaptations to the AIDS epidemic provided an opportunity to explore the congruence of data collected using a self-administered questionnaire with data from an unstructured face-to-face interview designed to facilitate report of sexual risk behavior. We examined (i) the concordance of questionnaire and unstructured interview risk ratings when the two data sources are scored for the same 16 sexual behaviors; (ii) the concordance of questionnaire ratings and ratings obtained when all information on recent sexual practices available from the unstructured interview is considered; (iii) the relationship between serostatus and both concordance patterns between methods; and (iv) the difference by serostatus of reported risk level within method. Riskiest behaviors were reported on the questionnaire for all serostatus groups. Riskier behaviors were more likely to be reported on the questionnaire while more characteristic, safer behaviors were discussed in interview, regardless of HIV serostatus. The advantages of a combined methods--questionnaire/interview--strategy for sexual practice research are discussed.