Peer support services are increasingly being integrated in programmes for the prevention of mother-to-child HIV transmission (PMTCT). We aimed to evaluate the effect of a peer-mother interactive programme on PMTCT outcomes among pregnant women on anti-retroviral treatment (ART) in routine healthcare in Dar es Salaam, Tanzania. Twenty-three health facilities were cluster-randomized to a peer-mother intervention and 24 to a control arm. We trained 92 ART experienced women with HIV to offer peer education, adherence and psychosocial support to women enrolling in PMTCT care at the intervention facilities. All pregnant women who enrolled in PMTCT care at the 47 facilities from 1st January 2018 to 31st December 2019 were identified and followed up to 31st July 2021. The primary outcome was time to ART attrition (no show >90 days since the scheduled appointment, excluding transfers) and any difference in one-year retention in PMTCT and ART care between intervention and control facilities. Secondary outcomes were maternal viral suppression (<400 viral copies/mL) and mother-to-child HIV transmission (MTCT) by ≥12 months post-partum. Analyses were done using Kaplan Meier and Cox regression (ART retention/attrition), generalized estimating equations (viral suppression) and random effects logistic regression (MTCT); reporting rates, proportions and 95% confidence intervals (CI). There were 1957 women in the peer-mother and 1384 in the control facilities who enrolled in routine PMTCT care during 2018–2019 and were followed for a median [interquartile range (IQR)] of 23 [10, 31] months. Women in both groups had similar median age of 30 [IQR 25, 35] years, but differed slightly with regard to proportions in the third pregnancy trimester (14% versus 19%); advanced HIV (22% versus 27%); and ART naïve (55% versus 47%). Peer-mother facilities had a significantly lower attrition rate per 1000 person months (95%CI) of 14 (13, 16) versus 18 (16, 19) and significantly higher one-year ART retention (95%CI) of 78% (76, 80) versus 74% (71, 76) in un-adjusted analyses, however in adjusted analyses the effect size was not statistically significant [adjusted hazard ratio of attrition (95%CI) = 0.85 (0.67, 1.08)]. Viral suppression (95%CI) was similar in both groups [92% (91, 93) versus 91% (90, 92)], but significantly higher among ART naïve women in peer-mother [91% (89, 92)] versus control [88% (86, 90)] facilities. MTCT (95%CI) was similar in both groups [2.2% (1.4, 3.4) versus 1.5% (0.7, 2.8)]. In conclusion, we learned that integration of peer-mother services in routine PMTCT care improved ART retention among all women and viral suppression among ART naïve women but had no significant influence on MTCT.