BACKGROUND: Infants of human immunodeficiency virus (HIV)-infected women have high mortality, but the immunologic integrity and protection afforded by the breast milk of HIV-infected women is unknown. METHODS: We determined morbidity and mortality by 24 months among breast-fed infants of 588 HIV-infected and 137 HIV-uninfected women followed-up in a clinical trial in Botswana. A matched case-control study compared clinical, behavioral, and breast milk immunologic parameters among 120 HIV-infected women by infant outcome. Breast milk factors were also compared between HIV-infected and HIV-uninfected women. RESULTS: Twenty-four-month mortality was 29.5% among HIV-infected infants, 6.7% among HIV-exposed uninfected infants, and 1.6% among HIV-unexposed infants. No differences were detected in breast milk immunologic profiles of HIV-infected women whose infants were either ill or well. Discontinuation of breast-feeding was the strongest predictor of illness (P<.001). Levels in breast milk of pathogen-specific immunoglobulin (Ig) G and IgA to Haemophilus influenzae, Campylobacter jejuni, Helicobacter pylori, Streptococcus pneumoniae, and innate immune factors were not lower among HIV-infected women than among HIV-uninfected women. CONCLUSIONS: Mortality was higher among HIV-infected and HIV-exposed infants than among HIV-unexposed infants. However, the immunologic profiles of breast milk among HIV-infected women were intact, and discontinuation of breast-feeding was the primary risk for infant morbidity. Thus, the breast milk of HIV-infected women may confer protection against common infant pathogens. TRIAL REGISTRATION: (ClinicalTrials.Gov) identifiers: NCT00197691 and NCT00197652.