The youths involved in the study were admitted to a detention center between 1994 and 1996. All were administered a standardized psychosocial screening assessment while in secure care, and they were released from secure care at least 6 months before the end of 1997. The assessment items used in this study were part of the screening instrument. Items included self-reports about how often various substances were used during the last month before being placed in secure care. The bivariate associations between the substance use variables (assessed while the juveniles were in secure care) and later recidivism status (6 months after release) were examined by using product-moment correlations and x square tests. A total of 447 juveniles were included in the recidivism analyses. Any new offending within 6 months was recorded. The study found that neither the use of a substance in general nor the use of any particular substance at the time of the last offense consistently predicted recidivism, nor did self-reported problems with substance use. Although marijuana was the most widely used substance, there was no support for using it as a marker for increased risk of reoffending. Recent alcohol and inhalant use were apparently the substances most consistently predictive of recidivism across various offending types and demographic subgroups. Both alcohol and inhalants are central nervous system depressants and have disinhibiting effects. The use of inhalants might also be a marker for less maturity and poorer self-control. An index of recent alcohol and inhalant use, although predictively too weak to use on its own, might be a useful addition to a broader recidivism risk assessment. 5 tables and 27 references