The aim of this population-based survey was to compare the prevalence of selected risk behaviors between students present or absent on the day of a school-based survey. The study population was a representative sample of all students of secondary schools in the Seychelles (Indian Ocean). Students absent on the day of the survey were traced and requested to complete the same self-administered questionnaire as did present students. Self-reported consumption of cigarettes, alcohol, and marijuana were measured. Of the sample of 1453 eligible students aged 11 to 17 years, 1321 "present students" completed the survey (90.9% participation), 11 refused to answer all questions, and 121 were not present at school. We could trace 105 of the 121 students not present at school on the survey day ("absent students"), and all of them completed the questionnaire over the next 4 weeks. The prevalence of risk behaviors was significantly higher in absent than present students for current smoking and drinking. Inclusion of data from the absent students resulted in a relative increase in the prevalence of the considered behaviors by 3% to 8% as compared to data based on present students only. In conclusion, the prevalence of risk behaviors was higher in absent than present students. Adjusting for data of absent students increased the prevalence estimates in the base population.