Background Retention is critical in HIV prevention programs and clinical research. We studied retention in the three modeled scenarios of primary prevention programs, cohort studies and clinical trials to identify predictors of retention. Methodology/Principal Findings Men attending Sexually Transmitted Infection (STI) clinics (n = 10, 801) were followed in a cohort study spanning over a ten year period (1993–2002) in Pune, India. Using pre-set definitions, cases with optimal retention in prevention program (n = 1286), cohort study (n = 940) and clinical trial (n = 896) were identified from this cohort. Equal number of controls matched for age and period of enrollment were selected. A case control analysis using conditional logistic regression was performed. Being employed was a predictor of lower retention in all the three modeled scenarios. Presence of genital ulcer disease (GUD), history of commercial sex work and living away from the family were predictors of lower retention in primary prevention, cohort study and clinical trial models respectively. Alcohol consumption predicted lower retention in cohort study and clinical trial models. Married monogamous men were less likely to be retained in the primary prevention and cohort study models. Conclusions/Significance Predicting potential drop-outs among the beneficiaries or research participants at entry point in the prevention programs and research respectively is possible. Suitable interventions might help in optimizing retention. Customized counseling to prepare the clients properly may help in their retention.