In people with opioid use disorder (OUD), medications for OUD (MOUD) prevent overdoses, but their effects are not well-studied in people with HIV (PWH) and OUD. PWH have unique access to and patterns of care engagement. We estimated the effect of initiating MOUD on overdose risk in a clinical cohort of PWH engaged in HIV continuity care in Baltimore, Maryland. MOUD was defined as self-reported methadone use or prescribed or dispensed buprenorphine. Overdose (due to any drug) was defined as any of: overdose as coded on the death certificate, an inpatient or outpatient overdose diagnosis, or self-report of ≥1 overdose in the prior 6 months on a patient-reported outcome survey asked regularly at clinic visits. Following index OUD diagnoses from 1 January 2019 to 31 March 2025, we cloned observations and assigned each to one of two treatment strategies: 1) initiate MOUD within a month or 2) do not initiate MOUD over 5 years of follow-up. We censored clones when participants stopped following the assigned treatment strategy. We adjusted for selection bias due to protocol violations using stabilized inverse probability of censoring weights estimated conditional on a comprehensive set of demographic, clinical, and social covariates. We fit weighted survival curves and calculated the risk difference. 95% confidence intervals (CI) were based on percentiles of the distribution of 500 estimates from bootstrap resamples. We included 1,280 incident OUD diagnoses from 365 PWH (57% Male, 83% Black, median age 57 years). The overall incidence rate was 7.7 overdoses per 100 person-years (95% CI: 6.8, 8.9). In the weighted analysis, initiating MOUD within a month of OUD diagnosis was associated with an 8.4% (95% CI: -1.7, 17.7) lower 5-year risk of overdose. Timely initiation of MOUD among PWH and OUD reduces the risk of overdose. More research is needed to improve MOUD uptake and persistence in this patient population to address the public health burden of the opioid epidemic.