Propositions underlying the study are derived from a review of three research traditions: the risk-prediction tradition, ecological/areal research on delinquency, and studies of neighborhood and community dynamics. The study obtained data on criminal history, current offense, social history, demographic characteristics, and performance after release for 500 offenders released from State institutions to any of 90 randomly sampled Baltimore neighborhoods over 2 years. Data were collected in June and July of 1981. Neighborhood characteristics were determined through onsite assessments using a checklist (building types, appearance, and social climate). Recidivism measurements included crime seriousness, time at risk, and time free. Analysis of covariance (by regression) was used to identify the unique contributions to recidivism of time available in which to recidivate, offender and environmental characteristics, and the interactions of persons and places. The study did increase recidivism predictive power by including environmental characteristics. The improvement was primarily due to the interaction of environmental and offender characteristics. An interactionist approach is proposed for future recidivism research. 4 tables and 18 footnotes.