Dynamic hepatic scintigraphy was performed in 49 patients with established cirrhosis, using intravenous 99Tcm-pertechnetate and 99Tcm-sulphur colloid in a prospective study of its predictive value. There was a close correlation between the hepatic perfusion index (reflecting the ratio of arterial to total hepatic blood flow) obtained with pertechnetate (HPI-P) and with sulphur colloid (HPI-C) (r = 0.775; p less than 0.0001), and both indices correlated with disease severity (HPI-P p less than 0.0001; HPI-C p less than 0.01). HPI-P was significantly increased in patients who died, in patients with varices and in those with hepatic encephalopathy. HPI-C was significantly increased in patients with varices, in patients with hepatic encephalopathy and in those who had bled from varices. Neither HPI-P nor HPI-C was able accurately to predict the development of complications during the follow-up period. The trapping index (TI), reflecting a combination of hepatic extraction efficiency, degree of intrahepatic shunting and extrahepatic extraction of colloid, was significantly impaired in patients who died and in those with ascites, varices and/or variceal bleeding, but not in patients with hepatic encephalopathy. The trapping index correlated with disease severity, as did the computer-derived spleen-liver ratio (S-L ratio). Neither TI nor S-L ratio was able to predict the development of complications. The clearance rate constant of colloid from peripheral blood, the uptake rate constants for liver and spleen, and splenic volume were all found to be unhelpful as indicators of disease severity or as predictors of complications. While perfusion indices derived by dynamic hepatic scintigraphy reflect the severity of the underlying liver disease, their determination on a single occasion appears to offer no benefit in predicting the likelihood of major complications.