The predictive value of laboratory results and renal histologic data was examined in 102 patients upon entry into prospective, randomized, therapeutic trials of lupus nephritis. Three clinical features at the time of entry into the study were individually associated with increased rates of renal failure: age less than 24 years, male gender, and an elevated serum creatinine level. Subjects with diffuse proliferative or membranoproliferative glomerulonephritis were at a modest but significantly increased risk for the development of end-stage renal disease compared with patients with other classes of lupus nephritis. Semiquantitative scores of histologic features (specified by activity and chronicity indexes) identified subgroups of patients with comparatively high renal failure rates. To address the controversial issue of whether renal histologic data significantly improve the outcome predictions in patients with lupus nephritis, multivariate survival models were generated, permitting simultaneous consideration of multiple prognostic factors. Outcome predictions based on the strongest clinical predictors (age, sex, and serum creatinine level) were significantly enhanced by the addition of activity and chronicity indexes. Only age and chronicity index contributed significantly to the five-variable model and together constituted a two-variable model, the predictions of which were similar to observed outcomes. In the context of the highly significant prognostic indicators (age and chronicity index), immunosuppressive agents appeared to provide a slight therapeutic advantage over oral corticosteroids alone.