These milieus usually have high rates of absolute poverty and relative economic deprivation, as well as weak social institutions. The persistent findings of delinquent peer contributions to delinquency have yet to be tested under conditions where social class and milieu effects are controlled. There is little empirical evidence to determine how adolescents in high-crime neighborhoods avoid delinquency despite frequent contact with delinquent peers. The differences between violent delinquents and other youths from comparable neighborhoods are little understood. This study contrasts a sample of chronically violent male juvenile offenders with the general male adolescent populations (students and school dropouts) from inner-city neighborhoods in four cities. Violent delinquents differ from other male adolescents in inner cities in their attachments to school, their perceptions of school safety, their associations with officially delinquent peers, their perceptions of weak maternal authority, and the extent to which they have been victims of crime. Peer delinquency and drug 'problems' predict the prevalence of three delinquency offense types for both violent offenders and neighborhood youths. Among violent delinquents, there appear to be different explanatory patterns, with one type better described by internal controls (locus of control), a developmental measure. Overall, there is strong support for integrated theory including control and learning components, and similar associations exist among inner-city youths as in the general adolescent male population. Despite the generally elevated rates of delinquency in inner cities, the explanations of serious and violent delinquency appear the same when subjects are sampled at the extremes of the distribution of behavior. (Publisher abstract)