Ideas proposed by labeling theorists in criminology have been the focus of considerable controversy. However, these ideas have often not been systematically formulated and tested. In order to go beyond "sensitizing concepts," it is necessary to restate the basic propositions in testable terms. In this paper, labeling and attribution perspectives were combined and labeling hypotheses were derived from attribution theory. These hypotheses were tested on a population of institutionalized juveniles in Washington State correctional facilities. We were interested in whether the age, sex, or race of a child activates stereotypes and biases staff attributions about the child. The results suggest that these characteristics do not affect certain staff perceptions of the child.