Longitudinal data were collected for two cohorts of families living in high-crime areas of the medium-sized metropolitan area of Eugene-Springfield (Oregon). Randomly selecting 6 of the top 10 schools generated a sample of 102 for the first cohort. The second cohort included a sample of 104 families obtained from 7 schools randomly selected from the year's list of the 10 schools with the highest delinquency rates. Multiagent/multimethod definitions for both family process and antisocial trait concepts defined latent constructs, and official records of police contacts were collected over a 5-year period. Data for hazard rates showed that risk of police arrest varied significantly as a function of the child's age. A continuous time regression analysis indicated that most individual family process constructs and the antisocial trait construct were significantly related to an altered risk of police contact. Multiple regression analysis demonstrated a similar effect in predicting age at onset. In keeping with the hypotheses, with all variables entered simultaneously, contributions of family process constructs were nonsignificant; only contributions for the trait score and social disadvantage remained significant. Thus, more antisocial children were at greater risk for early police arrest. 47 references, 5 tables, and 2 figures