Of 1,115 respondents, 87 percent completed the two-session HIV prevention intervention; of these, 69 percent completed the 6-month data collection. Factors related to completing the intervention were different from those related to completing the 6-month interview. Completing the intervention was associated with HIV risk behavior and with having a negative result on the project's HIV test. Research completion was related to indicators of a stable lifestyle. Individuals at highest risk were less likely to complete the data collection and were thus under- represented in the assessment of intervention efficacy. Results underscored the importance of measuring attrition from both the intervention and the research data collection in order to assess effects of selection bias on evaluation results. Implications of the results for HIV prevention program planners and evaluators are discussed. 22 references and 4 tables