Objective To evaluate whether the use of immunological markers in addition to CD4 + lymphocyte count can improve the prediction of the probability of developing AIDS within a given period. Design and setting Prospective multicentre cohort study of homosexual men. Patients A total of 447 HIV-positive homosexual men followed prospectively at 6-month intervals (median time of observation, 47 months). Methods Estimation of AIDS-free time using lifetable plots by Cutler and Ederer and Weibull parametric models. A stepwise multivariate regression analysis was used to calculate the optimal combination of the parameters studied. Results In general CD4 + lymphocyte counts are most important for the prediction of AIDS-free time. The use of serum levels of β2-microglobulin (β2M), immunoglobulin A (IgA) and erythrocyte sedimentation rate (ESR) can significantly improve the predictive value of CD4+ lymphocyte counts. However, the usefulness of these parameters depends on the stage of HIV disease. In patients with a CD4+ lymphocyte count > 500 x 106/l, only IgA level had a significant predictive value; none of the other parameters significantly improved the model. In patients with a CD4+ lymphocyte count < 500 x 106/l, the absolute number of CD4+ cells itself was the most important single predictive parameter, but the prediction of AIDS was significantly improved by the addition of the other parameters investigated. The most powerful combination of parameters in this group was CD4 + count, β2M and ESR. Conclusion Determination of serum IgA, β2M and ESR in addition to CD4 + lymphocyte count may aid the choice of specific therapeutic regimens or systems of care for HIV-positive individuals.