BACKGROUND: Increased global access and use of antiretroviral therapy (ART) for human immunodeficiency virus (HIV) has been postulated to undermine HIV prevention efforts by changing individual risk-taking behavior. This review aims to determine whether ART use is associated with changes in sexual or injecting risk-taking behavior or diagnosis of sexually transmitted infections (STIs). METHODS: A systematic review and meta-analysis was conducted of HIV-seropositive participants receiving ART compared with no ART use in experimental or observational studies. Primary outcomes included (1) any unprotected sexual intercourse, (2) STI diagnoses, and (3) any unsafe injecting behavior. RESULTS: Fifty-eight studies met the selection criteria. Fifty-six studies containing 32 857 participants reported unprotected sex; 11 studies containing 16 138 participants reported STI diagnoses; and 4 studies containing 1600 participants reported unsafe injecting behavior. All included studies were observational. Unprotected sex was lower in participants receiving ART than in those not receiving ART (odds ratio [OR], 0.73; 95% confidence interval [CI], .64-.83; P < .001; heterogeneity I(2) = 79%) in both high-income (n = 38) and low-/middle-income country (n = 18) settings, without any evidence of publication bias. STI diagnoses were also lower among individuals on ART (OR, 0.58; 95% CI, .33-1.01; P = .053; I(2) = 92%); however, there was no difference in injecting risk-taking behavior with antiretroviral use (OR, 0.90; 95% CI, .60-1.35; P = .6; I(2) = 0%). CONCLUSIONS: Despite concerns that use of ART might increase sexual or injecting risk-taking, available research suggests that unprotected sex is reduced among HIV-infected individuals on treatment. The reasons for this are not yet clear, although self-selection and mutually reinforcing effects of HIV treatment and prevention messages among people on ART are likely.