Cigarette smoking was measured in a naive tenth grade population under conditions expected to influence the student's willingness to admit smoking. All students were tested for smoking both by questionnaire and by expired-air carbon monoxide assessment. The carbon monoxide data were used to test the equivalence of the study groups and to partition the sample into smokers and nonsmokers. Of the smokers those who were advised in advance of the biological test were twice as likely to admit cigarette use in the past week compared to those who were advised of the testing procedure only after they had completed their questionnaire. A live explanation and demonstration of the biological testing procedure proved as effective as a videotaped message. These data support earlier reports of the 'bogus pipeline' effect. Several methodological issues are discussed which may explain previous failures to replicate this finding.