Objective: Unprotected sexual behavior, needle sharing, and a prison history are major correlates of hepatitis B Virus (HBV). These risk factors are common among homeless people who also have elevated rates of HBV. We examine whether these behaviors were associated with completion or loss to follow-up of the most intensive and successful condition of a 3-arm HBV vaccination intervention. Significant results would imply that those most in need are the least compliant. Contributions of baseline demographics, physical health, psychosocial variables, and health beliefs were also assessed. Design: Three-hundred thirty- one adults from Los Angeles' Skid Row were assigned to nurse-case-managed sessions with hepatitis education, incentives, and tracking. Successive predictive structural equation models assessed the amount of variance accounted for by the risk variables, demographics, and the health-related variables. Main Outcome Measures: (1) Completion of 3 injections by 6 months; and (2) loss to a 6-month follow-up questionnaire. Results: The 3 risk factors explained 2% of the variance in completion and 1% of the variance in loss. Adding the other variables increased the variance explained to 14% for completion and 13% for loss. African American ethnicity, positive coping, social support, poorer health, no prison history, and greater efficacy significantly predicted completion. White ethnicity, less social support, better health, and less intention to complete predicted participant loss. Conclusion: The program was not strongly rejected differentially as a function of preexisting hepatitis B risk behaviors. Programs designed for homeless people should include malleable psychosocial and health belief model variables. These aspects of the lives of homeless people provide leverage points for future interventions. © 2010 American Psychological Association.