To analyze the long-term survival factors associated with HIV infection, a prospective follow-up study of 165 HIV-infected patients was performed after a clinical, nutritional, and biological evaluation. Survival rate could be determined in 129 patients after a follow-up of 42 mo before the use of protease inhibitors. After univariate analysis, multivariate analysis was performed with the Cox regression proportional-hazard model. Survival curves were calculated and compared with the Kaplan, Meier, and log-rank tests. The study also analyzed the factors associated with impaired nutritional status at the beginning of the study and their effects on the long-term follow-up. Factors that could explain body weight loss before the study were the level of intakes, resting energy expenditure, chronic diarrhea, and the number of previous opportunistic infections. In the long-term follow-up, univariate analysis showed that nutritional status could be separated into four classes of body weight loss (BWL) by degree of loss (BWL < or = 5%, 5% < BWL < or = 10%, 10% < BWL < or = 20%, BWL > 20%); lean body mass (adjusted to height), body cell mass, CD4 count, albumin, prealbumin, and C-reactive protein (CRP) were all significant predictors. Age, stage of disease, number of previous opportunistic infections, and antiviral therapies were not associated with a change in survival. With the multivariate model, only CD4 counts, lean body mass/height squared, and CRP remained significant independent predictors of survival after controlling for other factors.