Hepatotoxicity may complicate therapy with methotrexate in patients with rheumatoid arthritis. Prevention of cirrhosis may depend upon early identification of liver damage, usually accomplished by serial biopsy. To determine the adequacy of noninvasive methods for identifying hepatotoxicity, 22 sets of data were obtained in patients undergoing therapy with methotrexate for rheumatoid arthritis. Comparisons were made between liver biopsy, hepatocellular enzymes and two noninvasive radioisotopic methods that have been shown to be abnormal in hepatocellular disease: the rate constant of excretion of the 14C-aminopyrine and the time from injection to peak hepatic activity of 99mTc-diisopropylimidodiacetic acid. The hepatocellular enzymes and the time-to-peak-activity of diisopropylimidodiacetic acid were not useful predictors of methotrexate-induced hepatotoxicity. The aminopyrine breath test was abnormal in approximately half the patients with hepatotoxicity but showed poor specificity. Noninvasive methods remain inferior to biopsy for the detection of mild to moderate methotrexate-induced hepatotoxicity in patients with rheumatoid arthritis.