Longitudinal cohort studies of HIV and substance use disorders play an important role in understanding these conditions, but high rates of attrition can threaten their integrity and generalizability. This study aimed to identify factors associated with attrition in a 5-year observational cohort study of 469 individuals with and without HIV infection and methamphetamine (MA) dependence. Rates of attrition in our four study groups were approximately 24% in HIV-MA-, 15% in HIV+MA-, 56% in HIV-MA+, and 47% in HIV+MA+ individuals. Predictors of attrition in the overall cohort included history of MA, alcohol, and other substance dependence, learning impairment, reduced cognitive reserve, and independence in activities of daily living (all ps < 0.05), but varied somewhat by clinical group. Of particular note, enrollment in a neuroimaging sub-study was associated with significantly boosted rates of retention in the MA groups. Results from this investigation highlight the complexity of the clinical factors that influence retention in cohort studies of HIV-infected MA users and might guide the development and implementation of targeted retention efforts.