The principal value of the renal biopsy in patients with SLE is as a therapeutic guide. Although semiquantitative indices of nephron loss (chronicity = CI) and acute potentially reversible inflammation (activity = AI) are reported by some to have separate values from traditional classifications of glomerular pathology as predictors of outcome and therapeutic guides, this point remains controversial. We have tested the predictive value of the AI and CI in a large group of patients with severe lupus glomerulonephritis (SLE GN) and a mean follow-up of 281 weeks +/- 116 (mean +/- SD). A total of 86 patients entered into the study of plasmapheresis in severe SLE GN by the Lupus Nephritis Collaborative Study Group, and long-term follow-up was available in 83. The predictive value of the AI and the CI was described over the entire range of cut-off points by the method of receiver operator characteristics (ROC). ROC analysis demonstrated that there was no level of either AI or CI that predicted the outcome of death or renal failure with sufficient sensitivity and specificity to be useful in the individual patient. The CI signifies renal damage and nephron loss, whereas the AI describes potentially reversible pathology. Neither the CI nor the AI taken by itself predicts individual outcomes of renal failure or death in patients with aggressively treated SLE GN. Since the indices fail to identify the patient whose disease will progress to renal failure, they are both insufficient as therapeutic guides and add little to the management of the patient with severe SLE GN.