This retrospective case study assessed the influence of item non-response error on subsequent response to questionnaire items assessing adolescent alcohol and marijuana use. Post-hoc analyses were conducted on survey results obtained from 4,371 7th to 12th grade students in Ohio in 2005. A skip pattern design in a conventional questionnaire appeared to increase nonresponse error rates at first introduction. In previous cigarette users, these errors were associated with under-reporting of alcohol and marijuana use on subsequent questions. These effects were less prevalent in high-achieving students. Skip patterns can contribute to substantial underestimation of alcohol and marijuana use in adolescents. (Contains 5 tables.)