A study of all boys who appeared in Cook County Juvenile Court (Illinois) charged with delinquency in 1928 showed that 8 out of 10 boys accused of delinquency were alleged to have committed their offenses in the company of 1 or more companions. More modern evidence is available on the predominance of groups as a distinctive aspect of adolescent criminality; data from the National Crime Panel, from the Vera Institute of Justice (New York), and from the Los Angeles Juvenile Court (California) exhibit similar characteristics to the 1928 Cook County data. However, attempts to use existing aggregate data on offenses, arrests, and punishments are confounded by the overlapping jurisdictions of juvenile and criminal courts. The problem is illustrated by examining common methods of estimating the risks of punishment and apprehension that are used to measure the credibility of threats in deterrence studies. Drawbacks in current approaches to measuring the incapacitation impact of incarceration, modeling patterns of criminal behavior, comprehending 'criminal' careers, and determining appropriate sanctions for youth crime also are analyzed. The article concludes that attempts to reform sentencing practices in the juvenile court, especially efforts to lead sanctioning models away from the jurisprudence of treatment and toward concepts of making the punishment fit the crime, will find the myriad problems of sanctioning the adolescent accomplice very close to the top of any sensible priority list for deliberation. Moreover, the special character of adolescent criminality requires special approaches. More generally, those who regulate particular forms of human behavior, or study the effects of regulation, abstract themselves from the knowledge base of other social and behavioral sciences only at great cost. Tables and 42 footnotes are provided.