Although studied for most of this century there is still no reliable model of severe decompensated micronodular cirrhosis that can be predictably produced in reasonable quantity. The current most successful model, inhalation of carbon tetrachloride vapor in the phenobarbitone-induced rat, has a low yield of severe cirrhosis and a high death rate because there is no way to determine both the variation in response to carbon tetrachloride and the maintenance of a constant critical level of liver damage. A new approach to this old problem is described in which both variation and level of critical damage are monitored by the daily weight change of the rat in response to intragastric carbon tetrachloride given during light halothane/oxygen anesthesia; the response each time being used to calibrate the subsequent dose of carbon tetrachloride to fit the individual rat. The method is effective in producing cirrhosis with ascites in about 75% of rats after 8-10 doses of carbon tetrachloride.