Objective: To test the hypothesis that modified perinatal mortality, early neonatal mortality, and other measures of perinatal mortality are lower with facility births than with home births among deliveries conducted by traditional birth attendants (TBAs) or nurse-midwives. Method: This population-based observational study used data collected prospectively for home and facility deliveries conducted by TBAs and nurse-midwives in 13 rural communities in Zambia between September 1, 2009, and December 31, 2015. Results: We enrolled 48 956 pregnant women. In adjusted analysis, modified perinatal mortality (adjusted odds ratio [aOR] 0.63, 95% confidence interval [CI] 0.45–0.88), early neonatal mortality (0.48, 0.33–0.69), and fresh stillbirth/day-1 neonatal mortality (0.55, 0.38–0.80) were lower among home deliveries than among facility deliveries conducted by TBAs, but did not differ among deliveries conducted by nurse-midwives. Rates of fresh stillbirth did not differ between home and facility delivery by either TBAs (aOR 1.03, 95% CI 0.64–1.66) or nurse-midwives (1.19, 0.67–2.10). Conclusion: Our findings show significant reductions in modified perinatal mortality, early neonatal mortality, and fresh stillbirth/day-1 neonatal mortality among home deliveries done by TBAs. This may be explained by robust community structures built by our program and referral bias of complicated cases.