Despite the fact that obtaining accurate self-reports presents challenges for sensitive topics, investigators often employ them to estimate the prevalence of a variety of sensitive behavior. This study examined the effect of four data collection techniques (face-to-face, computer, randomized response technique, and item count technique) on estimates of the prevalence of five sensitive topics (stealing, anal intercourse, lying, marijuana use, and cheating). Each respondent answered all of the five sensitive items, so that item served as a repeated measure. Each respondent responded using the same data collection technique, so that technique served as an independent groups factor. Data were collected at four locations at a large university in the Midwest United States. Four experimenters, two males and two females, solicited respondents. Only those walking alone were solicited, with a random number (2 to 6 inclusive) determining which of the passersby were approached. Consistent with past research the randomized response technique yielded the highest overall prevalence estimates, although for the most sensitive items estimates were within sampling error of those found in the computer condition. Given the advantages in ease and efficiency associated with collecting self-report data with computers, relative to collecting self-report data using the randomized response technique, these results suggest that collecting sensitive information with computers provides considerable value.