The schools were chosen by the Chicago Public Schools administration as middle schools with serious gang problems. The self-report survey questionnaires sought information about attitudes, perceptions, associations, symbolic behaviors, and activities that could most directly indicate gang involvement or readiness to be further involved in gang activity. In addition, self-report and official measures of delinquency were used to generate a single interval-level measure of delinquency. The data were analyzed using scale sequence and regression analyses. Results suggested that different social processes operate in gang involvement for the two ethnic populations. In both sets of cross-sectional data, the fitting of linear structural models revealed gang involvement to be an effective post-hoc estimator of delinquency for these youth, whereas delinquency is not an effective estimator of gang involvement. Tables, notes, and 49 references