Society increasingly holds schools responsible for the effectiveness of health promotion activities, such as drug abuse prevention efforts funded through the federal Safe and Drug-Free Schools program. Consequently, school districts use student surveys as a method for assessing trends and evaluating effects of programs on behavior. Because cost and practical concerns often preclude consistent population-based school survey sampling, risk indicators can provide an essential tool in analyzing needs assessment and program evaluation data. In this paper, three risk measures associated with substance use were selected from among commonly used school surveys. These measures--truancy, grade point average, and recent sexual intercourse--were compared, using meta-analysis techniques, to assess the reliability of risk measures across different survey instruments, different communities, and different points in time. Truancy was judged superior, because of its strong predictive value, particularly among younger students, and because rates can be compared to school records to assess sampling validity over time.