BACKGROUND: We assessed the impact of point-of-care (PoC) test-and-treat at birth on clinical outcomes and viral suppression among HIV-positive infants in Mozambique and Tanzania. METHODS: This cluster-randomized trial allocated health facilities to intervention, providing PoC-testing and antiretroviral treatment (ART) at birth and week 4-8, or control, starting these at week 4-8. The primary outcome was proportions of clinical events (mortality, morbidity, retention, virological failure, toxicity) among HIV-positive infants at month-18. We estimated incidence rate ratios adjusted for timing of HIV-detection (aIRR) and reported viral suppression <1000 copies/mL. FINDINGS: Among 6602 neonates enrolled October 2019-September 2021, 125 were diagnosed HIV-positive by week 12. In the intervention arm, 38/69 (55.1%) were diagnosed at birth with 35 initiating ART within two days. In the control arm, 27/56 (48.2%) were retrospectively detected HIV-positive at birth, of whom 6/56 (10.7%) died or were lost to follow-up before testing. Median age at ART initiation was 6 (intervention) versus 33 days (control). Birth test-and-treat was not associated with a significant reduction in clinical outcomes up to month-18 [53 (76.8%) versus 48 (85.7%); aIRR 0.857; 95% CI 0.505-1.492], but showed a 68% relative reduction in 6-month mortality. Viral suppression was poor overall, but improved in the intervention group at month 18 (65.7% versus 29.6%; p=0.005). INTERPRETATION: PoC test-and-treat at birth is feasible in resource-poor settings and resulted in clinically-relevant reduction of early infant mortality, though improved clinical outcomes were not sustained to month-18. Poor viral suppression may undermine early survival benefits, calling for better paediatric treatments and tailored adherence interventions.