To examine the prevalence, quantification, and factors that influence HIV in the cell-free compartment of breast milk, we performed reverse transcription polymerase chain reaction (RT-PCR) on samples obtained from HIV-1-infected study subjects. Virus was detected in 86 of 136 samples (63.2%) from 79 study subjects. HIV RNA quantity ranged from undetectable to 227,586 copies/ml. Prevalence and mean viral load were not affected by postnatal ages or maternal vitamin A supplementation. Among study subjects with multiple samples, breast milk viral load did not change at different postnatal ages. Breast milk viral load correlated positively with plasma viral load (r = 0.47; p =.005) and negatively with maternal CD4 count at entry to the study (r = -0.26; p =.02). Mothers of HIV-infected children had a higher proportion of detectable HIV RNA in their breast milk than mothers of uninfected children (p =.03) and higher mean log10 HIV RNA quantities (p =.04). In a multivariate logistic regression model, log10 HIV RNA quantity in breast milk was significantly associated with the risk of mother-child transmission (odds ratio [OR], 2.82; 95% confidence interval [CI], 1.22-6.51). Thus, prevention and treatment of opportunistic infections and of mastitis and early weaning may be important elements of a public health policy that is relevant to women in developing countries with HIV infection. Where available, antiretrovirals may also have an impact on opportunistic infections and mastitis.