A prospective randomized double-blind study was conducted to evaluate the efficacy of sodium benzoate in the treatment of acute portal-systemic encephalopathy. Seventy-four consecutive patients with cirrhosis or surgical portasystemic anastomosis and hepatic encephalopathy of less than 7 days duration were randomized to receive lactulose (dose adjusted for 2 or 3 semiformed stools/day) or sodium benzoate (5 gm twice daily). Assessment of response included mental status, asterixis, arterial ammonia level, electroencephalogram and number-connection test. Each was given a score between 0 and 4+. A portal-systemic encephalopathy index was calculated with these scores. Visual, auditory and somatosensory evoked potentials and a battery of psychometric tests for intelligence and memory were also performed to assess improvement. Thirty-eight patients received sodium benzoate; 36 took lactulose. Thirty patients (80%) receiving sodium benzoate and 29 (81%) receiving lactulose recovered; the remaining patients died. Improvement in portal-systemic encephalopathy parameters occurred in both treatment groups and was similar (p greater than 0.1). Electroencephalogram and evoked potentials were not as helpful as mental status in assessing of recovery. Psychometric test scores remained abnormal after recovery of mental status (21 to 42 days) and were probably too sensitive for monitoring of these patients. The incidence of side effects was similar in the two treatment groups. The cost of lactulose for one course of therapy was 30 times that of sodium benzoate. We conclude that sodium benzoate is a safe and effective alternative to lactulose in the treatment of acute portasystemic encephalopathy.