The short-term prognosis of lupus nephritis was evaluated by assessing serum creatinine 12 months after renal biopsy in 87 patients with lupus nephritis. On univariate analysis, significant clinical and laboratory predictors of this outcome included clinical signs of renal injury (serum creatinine, 24-hour urinary protein, prolonged renal disease, nephrotic syndrome, serum albumin), as well as thrombocytopenia, older age, and coexisting illness or hypertension at the time of biopsy. On renal biopsy, diffuse proliferative nephritis, higher activity, chronicity, or tubulointerstitial scores, or subendothelial or subepithelial electron dense deposits predicted a higher serum creatinine 12 months after biopsy. A clinical predictive model was developed which included as independent predictors serum creatinine, age, platelet count and 24-hour urinary protein. Any one of three biopsy variables added information to the clinical prediction model: a marked quantity of subendothelial deposits (p = 0.02), a higher activity index score (p = 0.02), or the presence of diffuse proliferative lupus nephritis (p = 0.05). However, the relative predictive accuracy of the clinical model did not improve with the addition of any of the biopsy variables. The value of renal biopsy in lupus nephritis is discussed based on the ability of biopsy information to confirm the prognosis, to add new predictive information for a group of subjects, and to improve predictive accuracy for individual patients.