The Addiction Severity Index (ASI) is an instrument widely used to assess the treatment problems of substance users. Its psychometric properties have been tested and found satisfactory for many types of substance abusers entering treatment. However, there are many other subgroups of substance users not in formal treatment, such as homeless substance users. While the ASI has been used with this subgroup, its psychometric properties remain questionable. This study examined the reliability and validity of the ASI in a sample of 98 homeless substance users awaiting temporary housing placement. Test-retest reliability found the ASI to have moderate to high reliability coefficients in each of the seven domains assessed. Both composite score and severity rating measures were found to be quite independent with low intercorrelations. Three of the seven ASI composite scores were tested for and found to have moderate concurrent validity: alcohol (r = .31 to .36), drug (r = .46), and psychiatric (r = .53 to .66). Composite score interitem correlations were .70 or greater in each of the domains except for employment (.50) and family (.52). These data suggest that, although there are some limitations in using the ASI with homeless substance users, it demonstrated acceptable reliability and validity.