The Internet has become an inevitable tool for collecting health and health behavior questionnaires. This study compared the feasibility, presence of score differences, and subjective evaluations by children between Internet and identical paper (asthma/fruit) questionnaires in elementary schools. A randomized crossover design was applied, with children starting with one administration mode before completing (5 minutes later) the other mode. Ten Dutch elementary schools with 270 school children (fifth grade, 10-12 years) were approached to participate. Response was 92%. The Internet mode had significantly less missing/nonunique answers than the paper mode (p less than 0.01). The completion times did not differ significantly between the Internet and the paper mode. Except for perceived self-efficacy to eat sufficient fruit (p less than 0.05), no differences in the asthma and fruit scores were found when comparing between the 2 modes. All variables showed strong intraclass correlation coefficients (166 greater than or equal to 0.64) between modes. Most items had good to very good agreement (kappa 0.61-0.95). The percentages for global and exact agreement ranged per item from 61.3 to 100. Most children preferred the Internet mode on "general preference" and "'ease of use" aspects but rated no preference on "understandability." The majority rated the evaluation aspects of the Internet mode positively. Thus, Internet administration of a health and health behavior questionnaire is feasible at elementary schools, gives comparable responses, and is well accepted and preferred by children compared to the paper version. Therefore, it is recommended to use Internet as a tool in health and health behavior research among children.