A secondary analysis of treatment files from the National Breakthrough Inhalant Program was conducted. Data included basic demographic information, a description of substance abuse problems, medical information, family background, and treatment information. Additionally, one of the researchers met with follow-up workers from home communities to obtain information about what happened to the youth after discharge from treatment. File information was used to develop a prediction model for relapse during follow-up. Seventy-four percent of the 78 youth tracked during follow-up relapsed after discharge from treatment. Many of the youth came from backgrounds marked by isolation, poverty, family violence, and substance abuse. The average age at which the youth started using solvents was 9.72 years. Gasoline was the most common inhalant used. Inhalant use was often accompanied by alcohol and drug abuse. A logistic regression model that predicted who would relapse indicated that youth who abused inhalants immediately before admission, those who were described as unmotivated in treatment, and those who were hospitalized during treatment had the greatest risk of relapsing during follow-up. Implications for policy are discussed. 5 tables and 26 references