A review of the research shows that there are substantial and consistent differences in drug use prevalence estimates between school-based surveys and household surveys. Significant differences in drug use prevalence estimates have also been observed between different school surveys. These findings suggest that respondents are under-reporting their drug use behaviors in an effort to conform to perceptions of social desirability. Over-reporting may also occur in an effort to project a more adult or deviant persona to researchers or peers. The problems of under- and over-reporting are likely to be accentuated in a school setting where both peers and teachers are prominent and visible. The reliability and accuracy drug use data obtained from school surveys becomes problematic when interventions are designed based on erroneous data. For example, researchers may design tobacco and alcohol cessation programming based on student data while ignoring interventions for heroin. The result is that young people just beginning to experiment with heroin and other more serious drugs do not have school-based interventions available to them because school programming has narrowly focused on tobacco and alcohol interventions. The findings of this review have indicated a critical need to gain a better understanding of the nature and extent of the problem of unreliable drug use data among young people in school settings. Several avenues to future research in this area are offered and include examining the possibility of increasing the voluntary nature of school-based surveys and the possibility of conducting classroom surveys without the presence of teachers. Taking into account all the difficulties inherent in collecting accurate drug use data from youth in school settings, the authors question whether schools are the most appropriate settings to collect such information. References