OBJECTIVE To develop a prognosis-based clinical staging system for infants infected with human immunodeficiency virus. METHODS Abstraction of data from medical records of 75 infected children. For each clinical finding present in infancy, the magnitude of the relative risk (RR) for early death was used to assign subjects to different clinical stages. RESULTS Stage IV (RR > 3) included subjects with Pneumocystis carinii pneumonia, other opportunistic infections, or encephalopathy. Stage III (RR, 2 to 3) included those with anemia, thrombocytopenia, hepatitis, fever, oral candidiasis, or one or more serious bacterial infections. Stage II included those with hepatomegaly, splenomegaly, failure to thrive, or persistent diarrhea, and stage I included those who had lymphadenopathy or were free of symptoms. When clinical staging was applied to the study population at ages as early as 6 months, survival curves were significantly different (IV vs III: p < 0.0005; III vs II + I: p < 0.005). CONCLUSIONS Clinical staging should be beneficial in advising parents about an infant's prognosis, therapeutic decision making, and stratification for clinical trials.