The aim of the present study was to examine the predictive accuracy of noninvasive clinical and biochemical variables associated with cirrhosis among patients with C282Y homozygous hemochromatosis. Sixteen clinical and laboratory variables were recorded at the time of diagnosis in 193 Canadian C282Y homozygous patients. All patients underwent percutaneous liver biopsy and 27 (14%) had biopsy specimen-proven cirrhosis. Prediction of cirrhosis was assessed first by univariate regression analysis. Variables significantly related to cirrhosis were then evaluated by stepwise linear multivariate regression. Receiver operating characteristic curve analysis of the most informative variables from multivariate analysis was then used to devise a clinically applicable index for the noninvasive prediction of cirrhosis. This index was then validated in 162 C282Y homozygous patients in France. Ferritin, blood platelets, and aspartate transaminase (AST) level were selected for the clinical index. The combination of ferritin levels of 1,000 microg/L or greater, platelet levels of 200 x 10(9)/L or less, and AST levels above the upper limit of normal led to a correct diagnosis of cirrhosis in 77% of Canadian patients. In the French patients, this led to a correct diagnosis of cirrhosis in 90%. In conclusion, in C282Y homozygous patients, a combination of easily measured laboratory variables (ferritin, platelets, AST) can be used to make the diagnosis of cirrhosis in approximately 81% of cases, reducing the need for liver biopsy.