Attrition is a concern to researchers because of implications regarding the potential for inaccurate study findings. Individuals with a substance abuse problem can be more challenging to retain in longitudinal research both because they are more difficult to track and because they are more likely to drop out than participants in other research studies. The goal of this study was to identify the factors that predicted attrition of St. Louis Target Cities participants enrolled in a longitudinal study. A higher attrition rate was anticipated for men and those that were unemployed, had not completed high school, did not own their own residence, had high substance abuse severity or a history of mental illness, or had a criminal record. A higher attrition rate was also expected for those that did not complete either of their first two early engagement interviews, listed fewer than three contacts, and did not attend treatment. Results show that participants were more likely to drop out if they were male, unmarried, unemployed, failed to complete high school, or did not own their own residence. This study failed to find that race or history of a criminal record influenced study drop out. Obtaining three or more contacts at the recruiting interview and completion of an early initial follow-up interview were associated with lessened study attrition. The study found that alcohol and drug severity were not related to attrition, and that participants that had previously received psychiatric treatment were three times more likely to be retained in the study than participants that had never received psychiatric treatment. This study generally supported the findings of previous research, and underscores the importance of early engagement in a research protocol as one way to minimize attrition. 2 tables, 15 references