When respondents answer paper-and-pencil (PP) questionnaires, they sometimes modify their responses to correspond to previously answered items. As a result, this response bias might artificially inflate the reliability of PP questionnaires. We compared the internal consistency of PP questionnaires to computerized questionnaires that presented a different number of items on a computer screen simultaneously. Study 1 showed that a PP questionnaire's internal consistency was higher than that of the same questionnaire presented on a computer screen with one, two or four questions per screen. Study 2 replicated these findings to show that internal consistency was also relatively high when all questions were shown on one screen. This suggests that the differences found in Study 1 were not due to the difference in presentation medium. Thus, this paper suggests that reliability measures of PP questionnaires might be inflated because of a response bias resulting from participants cross-checking their answers against ones given to previous questions. (Contains 1 table and 2 footnotes.)