Background: We report effectiveness of an HIV-prevention intervention delivered by community health workers (CHWs) in Mombasa, Kenya, to PLHIV who have not initiated or who have discontinued ART - an often difficult-to-reach population because they fall outside the ambit of health care and prevention services. Methods: A 2-arm cohort study assessed a structured risk-reduction intervention involving at least 4 oneto- one counseling sessions and personalized support. The control group received standard prevention services. CHWs recruited treatment-naïve people living with HIV (PLHIV) or those who had previously taken antiretroviral drugs. Data were analyzed using a Propensity Score Matched (PSM)-sample to control for baseline differences between the groups. Results: 634 PLHIV were recruited and followed for 6 months. Median age was 35 years, and 74.3% were female. Participants in the intervention group reported reduced risky sexual behaviors both at endline compared with baseline and compared with the control group. At endline, in the PSManalysis, participants in the intervention arm were less likely than participants in the control group to report unprotected sex with a spouse (Odds Ratio [OR]50.08, 95% confidence interval [CI]50.03-0.24), and they reported fewer unprotected sex acts (12.3% versus 46.0%, respectively; OR50.16, 95% CI50.09-0.29; P<0.001). Further, 92.4% of participants in the intervention group reported zero unsafe sex acts (with partners of negative or unknown HIV status) compared with 70.8% in the control group (P<0.001), and more participants in the intervention arm were receiving ART (34.3% versus 12.7%, respectively; P<0.001). Conclusion: CHWs effectively reached PLHIV who had never received or who had discontinued ART, and they delivered a risk-reduction intervention that led to declines in reported sexual risk behaviors, as well as to increases in ART uptake. A scaled-up intervention warrants consideration.