OBJECTIVE: Training midlevel providers (MLPs) to conduct surgical abortions and manage medical abortions has been proposed as a way to increase women's access to safe abortion. This paper reviews the evidence that compares the effectiveness and safety of abortion procedures administered by MLPs versus doctors. METHODS: A systematic search was conducted of published trials and comparison studies assessing the effectiveness and/or safety of abortion provided by MLPs compared to doctors. The Cochrane Central Register of Controlled Trials, EMBASE, MEDLINE, and Popline were searched. The primary outcomes of interest were: (1) incomplete or failed abortion; and (2) measures of safety (adverse events and complications) of abortion procedures administered by MLPs and doctors. Odds ratios (ORs) and their 95% confidence intervals (CIs) were calculated for each study. Data were synthesized in a narrative fashion. FINDINGS: Five studies were included in this review (n = 8539 women), comprising two randomized controlled trials (RCTs) (n = 3821) and three prospective cohort studies (n = 4718). In total, 4198 women underwent a procedure administered by an MLP, and 4341 women underwent a physician-administered procedure. Studies took place in the US, Nepal, South Africa, Vietnam, and India. Four studies used surgical abortion with maximum gestational ages ranging from 10 to 16+ weeks, while a medical abortion study had gestational ages up to 9 weeks. In RCTs, the effect estimates for incomplete or failed abortion for procedures performed by MLPs compared with doctors were OR = 2.00 (95% CI 0.85-4.68) for surgical abortion, and OR = 0.69 (95% CI 0.34-1.37) for medical abortion. Complications were rare among both provider types (1.2%-3.1%; OR = 1.80, 95% CI 0.83-3.90 for surgical abortions), and no deaths were reported. CONCLUSION: There were no statistical differences in incomplete abortion and complications for first trimester surgical and medical abortion up to 9 weeks performed by MLPs compared with physicians. Further studies are required to establish more precise effect estimates.