This study compares the effectiveness of audio computer-assisted self-interviewing (Audio-CASI) with face-to-face interviews and self-administered questionnaires in collecting sensitive information on risky sexual and other behaviors among young men in urban India. A randomized study design compared data collected from 900 male college students using the three data-collection approaches and from 600 young men residing in slums using Audio-CASI and face-to-face interviews. Among the college students, the reported prevalence of risky behaviors was generally higher for young men interviewed through the Audio-CASI approach than with face-to-face interviews; self-administered questionnaires failed to yield significantly higher estimates than face-to-face interviews. Among the slum residents, the results were more mixed; the Audio-CASI approach failed to yield consistently higher responses for many risky behaviors compared with the face-to-face interview mode. The results demonstrate that although Audio-CASI appears to yield higher estimates of risky behavior among college-educated, computer-literate populations of young men, the efficacy of this approach among less-educated and less computer-literate populations appears more doubtful.