Delinquent boys who had been committed to a training school were interviewed by a psychiatrist to determine drug abuse status. Crimes that were officially recorded were classified as person offenses or property offenses, and the number of offenses was determined from the juvenile record. Drug scores were given subjects for each of 13 drug categories according to the following scale: 0, no use of drug; 1 to 4 v, varying use frequencies increasing in stepwise fashion but not reaching abuse proportions; 5, use greater than 8 days per month and abuse as defined in DSM III. Subject's number of crimes against persons were correlated with subject's drug scores for each of 13 drug categories. Five drugs had significantly positive correlations: phencyclidine, barbiturates, cocaine, Valium, and amphetamines. Scores for alcohol, LSD, mescaline, volatile substances, codeine, marijuana, heroin, and morphine did not correlate significantly with the number of person offenses in the person offender group. Mean drug scores for 13 drug categories of all subjects who were violent person offenders were compared with drug scores of subjects who were property only offenders. The t tests for independent means revealed that drug scores were significantly higher in the person offending group for five drugs: marijuana, alcohol, cocaine, heroin, and LSD. A high drug score for marijuana was the most predictive of all drug scores for membership in the person offending group.