The objective of this study was to investigate the relationship of plasma HIV RNA load with survival and disease progression in HIV-infected children and to determine its correlation with cellular HIV DNA. Virus load (VL, HIV RNA copies/ml) was determined retrospectively by nucleic acid sequence-based amplification (NASBA) assay in 144 stored plasma samples between birth and 48 months in 50 children of whom 40 are alive (age range, 2-13 years). On the basis of clinical and immunologic status children were classified as rapid progressors (RPs), or nonrapid progressors (NRPs). Proviral HIV DNA quantitated by QC-PCR (quantitative competitive polymerase chain reaction) in 24 children was compared with plasma HIV RNA. At age <3 months, plasma VL <750,000 copies/ml was associated with significantly higher survival to age >2 years (p < or =0.01) compared with a VL of > or =750,000 copies/ml. Increasing mortality was observed with increasing plasma HIV RNA levels at ages 3-24 months and baseline VL of infants who died before age 24 months was significantly higher (p = 0.004) than baseline VL of those who survived beyond 24 months. Although baseline VL in infants classified as RPs was higher than that of NRPs, the difference was not statistically significant. Among surviving children 2-13 years of age, the baseline VL obtained at <24 months of age was not predictive of disease severity. Although no significant correlation was noted between plasma HIV RNA and proviral DNA, the concurrence of positive and negative results was >80%. We conclude that high plasma HIV RNA in infancy is associated with increased mortality.