We assessed the association between marijuana use and retention in HIV care through a retrospective cohort study of patients engaged in care at a large HIV clinic in 2011 and 2012. Two different retention outcomes were assessed: not meeting the Institute of Medicine's (IOM) retention definition (≥2 provider visits ≥90 days apart in a calendar year) and no-show visits. Any marijuana use and frequency of marijuana use were obtained from a substance use screening questionnaire administered at each clinic visit. Modified Poisson regression was used to estimate risk ratios and 95% confidence intervals for the association between marijuana use and retention outcomes. Marijuana use was reported by 17% of 1791 patients and 21% were not retained (IOM definition). Marijuana use was not associated with the IOM retention outcome, but was associated with missing the next scheduled appointment. A non-linear dose-response was observed for frequency of marijuana use and missed visits, with daily users having the highest risk compared to non-users. Daily marijuana use had a negative impact on HIV clinic attendance. Further research is needed to elucidate the mechanisms by which marijuana use affects this outcome to inform targeted interventions.