This study was aimed at estimating the maturity of the human immunodeficiency virus (HIV) epidemic in cohorts of injecting drug users with existing HIV infection at the time of first observation, and using this information to estimate the incubation period of acquired immunodeficiency syndrome (AIDS) in this population group. The method imputed the "missing" time from seroconversion to first observation for seroprevalent subjects in a combined cohort including injecting drug users from New York (n = 246) and Baltimore (n = 621). Imputation relied on a probability model predicting time since seroconversion on the basis of two markers of maturity of HIV infection: percent of CD4+ lymphocytes and platelet count. The model was developed from data observed in a combined cohort of subjects from New York (n = 24) and Baltimore (n = 112) who had incident HIV infections. The estimates of median time since seroconversion for the Baltimore and Bronx seroprevalent subcohorts were 28 and 39 months, respectively. The total time from seroconversion to AIDS was then estimated for the incident plus completed-prevalent cohorts using a modified version of the nonparametric product-limit method. The results showed that approximately 95% (95% confidence interval: 90 to 98%) of drug users remained AIDS-free 2 years after seroconversion; 83% (74-91%), 4 years after seroconversion; and 72% (61-82%), 6 years after seroconversion. Median time to AIDS was 10.2 years, with an estimated 95% confidence interval of 7.9 to 12.3 years. Consistency of these results with those derived from large cohorts of homosexual men indicate that the HIV incubation distributions for drug users and homosexual men are similar.