The RBA is a structured interview that takes between 30 and 45 minutes to complete; it addresses demographics, drug-risk behaviors, heterosexual and homosexual sex-risk behaviors, drug treatment experience, health conditions that might be related to HIV risk, and economic variables. In addition, there are several items related to arrest and incarceration experience. Participants in this reliability study were 229 paid volunteers recruited through targeted sampling plans developed at each of 11 sites. At Time 1, participants were administered the RBA, and urine and blood samples were collected for drug and HIV testing. At Time 2 (48 hours later), they were once again administered the RBA, and another urine sample was collected. All testing was completed during March 1992. The findings indicate reliability that varied from excellent to very poor. Specifically, reliability coefficients for the self-reported total amount of time participants had spent in jail or prison in their lifetimes had excellent reliability. Similarly, the dichotomized variable of no arrest versus any arrests in a lifetime had acceptable reliability. The question on the number of times in a lifetime that participants reported being arrested and charged/booked with a criminal offense showed moderate reliability; for recent behavior (30 days prior to the RBA interview), the reliability coefficients were very low. Future research should explore alternative RBA items that can produce reliable data on recent arrest history. 1 table and 20 references