The Women and Infants Transmission Study is a prospective natural history study that has been enrolling HIV-1-infected pregnant women and their infants since 1989. To evaluate the impact of different antiretroviral regimens on perinatal HIV-1 transmission at the population level. Prospective cohort study. Plasma HIV-1 RNA levels were serially measured in 1542 HIV-1-infected women with singleton live births between January 1990 and June 2000. HIV-1 status of the infant. HIV-1 transmission was 20.0% (95% confidence interval [CI], 16.1%-23.9%) for 396 women who not receiving prenatal antiretroviral therapy; 10.4% (95% CI, 8.2%-12.6%) for 710 receiving zidovudine monotherapy; 3.8% (95% CI, 1.1%-6.5%) for 186 receiving dual antiretroviral therapy with no or one highly active drug (Multi-ART); and 1.2% (95% CI, 0-2.5%) for 250 receiving highly active antiretroviral therapy (HAART). Transmission also varied by maternal delivery HIV RNA level: 1.0% for <400; 5.3% for 400 to 3499; 9.3% for 3500 to 9999; 14.7% for 10,000 to 29,999; and 23.4% for >30,000 copies/mL (p =.0001 for trend). The odds of transmission increased 2.4-fold (95% CI, 1.7-3.5) for every log10 increase in delivery viral load. In multivariate analyses adjusting for maternal viral load, duration of therapy, and other factors, the odds ratio for transmission for women receiving Multi-ART and HAART compared with those receiving ZDV monotherapy was 0.30 (95% CI, 0.09-1.02) and 0.27 (95% CI, 0.08-0.94), respectively. Levels of HIV-1 RNA at delivery and prenatal antiretroviral therapy were independently associated with transmission. The protective effect of therapy increased with the complexity and duration of the regimen. HAART was associated with the lowest rates of transmission.