Three experimental investigations of the effects of observers in a laboratory classroom are described. In each study, classroom observers were systematically introduced and withdrawn while additional observers collected data surreptitiously. These studies revealed little systematic effect on the behavior of children or their teacher. Further, the effects which were manifest (a) were not replicable across studies, and (b) did not exceed the number of significant effects expected on the basis of chance. In addition, teachers' evaluations of the effects of repeated observations in 74 regular classrooms provided no evidence of reactivity. Both methodological and ethical implications of these results are discussed.