To determine if beta 2-microglobulin (beta 2M) predicts death among HIV-infected African women. Nested case-control study. Kigali, Rwanda. Two hundred and five seroprevalent women known to be HIV-infected since 1986-1987; 67 of whom died of HIV disease (cases) and 138 were alive (controls) as of November 1991. In addition, 128 women who seroconverted between 1986 and 1991. HIV serology, clinical signs and symptoms of HIV disease, hematology variables, and beta 2M concentration. beta 2M concentration increased over time (P < 0.001) in the seroprevalent women and seroconvertors. The average rate of beta 2M increase in women who died was 0.5 compared with 0.3 mg/l/year in the vital, seroprevalent women (P = 0.07). The strongest independent predictors of death were the rate of change of beta 2M (mg/l/year) [odds ratio (OR), 3.4; 95% confidence interval (CI), 1.7-6.8] and baseline beta 2M concentration (mg/l) [OR, 1.6; 95% CI, 1.2-2.1]. The rate of death for women with beta 2M concentration > or = 7.0 mg/l and a rate of change of beta 2M > or = 0.4 mg/l/year was 7.3 times higher than for women with beta 2M concentration < 7.0 mg/l and a rate of change of beta 2M of < 0.4 mg/l/year (95% CI, 3.1-17.2). The estimated median time from seroconversion to death assuming a constant rate of change of beta 2M was 10.6 years (95% CI, 9.9-11.2) for this cohort of HIV-infected women. Elevated beta 2M and a high rate of beta 2M increase were strongly associated with mortality among HIV-infected African women. Based on survival estimates using the rate of change of beta 2M, HIV-infected African women have similar survival compared with HIV-infected adults in the United States.