Although sampling is a crucial component of research methodology, it has received little attention in intervention research with populations at risk for HIV infection. We review the challenges involved in sampling these populations for evaluating behavioral and social interventions. We assess the four strategies used for street and network sampling that have been reported in the HIV-intervention research literature and used because traditional probability sampling was not possible. The sampling strategies are: 1) targeted, 2) stratified, (3) time-space, and (4) respondent-driven. Although each has strengths and limitations in terms of its ability to produce valid results that enhance generalizability, the choice of a particular strategy depends on the goal of the study, characteristics of the target population, and the availability of resources and time for collecting and analyzing sampling-related data. Continued efforts are needed to improve the sampling strategies used in evaluation studies of HIV risk-reduction interventions.