Increased, s3infmetrical signals of varying intensity in the globus pallidi on T1-weighted (TlW) images, without corresponding signals on T2-weighted (T2W) images, have been reported previously in chronic hepatic failure. It has been suggested that these signals are characteristic of chronic hepatic encephalopathy. To test this hypothesis, we evaluated the relationship of magnetic resonance imaging (MRI) abnormalities with ammonia, albumin, bilirubin, prothrombin time, ascites, clinical encephalopathy, and neuropsychological tests in 46 patients (16 with alcohol-induced cirrhosis and 30 with non-alcohol-induced cirrhosis). T1W signal and cortical atrophy were graded by a neuroradiologist in a blinded fashion. Eleven patients had no T1W signal, 18 had minimal TlW signal, and 17 had high T1W signal. Twentyfive patients had no cortical atrophy, 14 had mild atrophy, and 7 had moderate atrophy. Cortical atrophy was noted more commonly in patients with alcohol-induced liver disease. The neuropsychological tests correlated significantly with albumin, prothrombin time, Child-Pugh's score, clinical encephalopathy, and ammonia. T1W signal and cortical atrophy did not correlate with the neuropsychological tests, clinical encephalopathy, ascites, albumin, prothrombin time, ammonia, or Child-Pugh's score. Patients with high T1W signal showed significant correlation with bilirubin (P < .005). This study suggests that globus pallidus signal in T1W images observed in liver disease are not indicators of chronic hepatic encephalopathy as previously assumed. (HEPATOL-