Positive consent was solicited from parents of 604 seventh grade students in four middle schools. Three hundred and fifty eight (59%) returned consents and completed a questionnaire under " bogus pipeline" conditions with saliva and air samples. Two weeks later both students with consent and those without were administered a second questionnaire without physiological measures. Comparison between consent and nonconsent students show significant differences in the smoking of cigarettes and marijuana, but no difference in the use of alcohol. Additional significant differences were found in exposure to smoking models, and level of education of both parents. The bias shown on significant dependent variables may adversely effect the generalizability of results of studies of adolescent drug use that depend upon positive parental consent.