This study evaluated perceptions of masked assessors to determine whether there are subtle differences in cocaine-exposed and unexposed children who might be identified by those interacting with children. As part of a longitudinal study, developmental assessors were masked to 163 4-year-old children's actual in utero cocaine exposure status and developmental history. After each battery, assessors documented their guesses of the child's cocaine exposure. Thirty-seven percent of the children who were exposed were misclassified as unexposed, whereas 74% of those unexposed were incorrectly classified as exposed. Although the sample did not differ on assessment scores when results were analyzed by actual cocaine exposure status ( >.3), children who did less well on assessments were more likely to be labeled by assessors as cocaine-exposed ( <.001). Results highlight the potential of stereotyping and negative attributions that might distort observations, both in unmasked studies of prenatal cocaine exposure and in clinical settings.