Cannabis use is prevalent among HIV-positive persons, but evidence regarding the impact of cannabis in HIV-positive persons is limited. We conducted a retrospective cohort study of HIV-positive adults initiating their first antiretroviral therapy (ART) regimen. A dedicated intake form assessed self-reported cannabis use in the preceding 7 days at each visit. The relationships between time-varying cannabis use and body mass index (BMI), CD4+ T-cell count, and HIV-1 RNA levels were assessed using random effects models adjusted for age, sex, race, and other reported substance use. 4290 patient-visits from 2008 to 2011 were available from 1010 patients. Overall, there were no statistically significant differences in CD4+ T-cell count and BMI across multiple adjusted models using different measures of cannabis use (ever use during the study period, any use, and number of times used in the preceding 7 days). Cannabis use by all three measures was associated with greater odds of having a detectable viral load at a given visit than no reported use (OR 2.02, 1.72, and 1.08, respectively; all adjusted p < 0.05). Self-reported cannabis use was not associated with changes in BMI or CD4+ T-cell count in ART-naïve HIV-positive persons starting treatment. However, reported cannabis use by multiple categories was associated with having a detectable HIV-1 RNA during the study period. Associations between cannabis use, adherence, and HIV-related outcomes merit further study.