Can one view deviance as a developmental process in which one type of deviant act leads to another? This paper proposes a number of criteria that would need to be met if there is such a process and applies them to data from records and retrospective interviews about the ages at which 13 kinds of childhood behaviors began. Results appear consistent with both a quantitative developmental process, i.e., one in which the probability of committing a new type of deviance is in part a function of the variety of acts previously committed, and a qualitative one, i.e., one in which having committed one particular type of deviant act makes more probable initiating another particular type of deviance thereafter.