Fourteen patients with cirrhosis and subclinical hepatic encephalopathy were randomised to treatment with lactitol or lactulose for a 2-month period during which they were monitored clinically, by electroencephalography and by manually administered and computer-based psychometric testing. Following a washout period of 4-6 weeks patients were crossed-over to treatment with the alternative sugar for a similar period of monitoring. None of the patients showed evidence of overt hepatic encephalopathy and only one showed slowing of the electroencephalogram mean cycle frequency at the onset of the trial. However, significant impairment was observed in the group as a whole in the performance of all three manually administered psychometric tests and in four of the ten computer-based test variables. No changes were observed in clinical status or in electroencephalogram mean cycle frequency during treatment with either lactitol or lactulose. However, psychometric performance improved consistently, and to the same degree, during treatment with both sugars. Patients required a mean of 26 g (range 8-36) of lactitol and 25 ml (10-60) of lactulose to achieve two semi-soft stools per day. The majority of patients complained of flatulence during treatment with both sugars but this tended to resolve with continued treatment. Diarrhoea developed in a small number of patients during both treatment periods but this was invariably dose-related. Patients were equally divided in their preference for the two sugars. Patients with subclinical hepatic encephalopathy benefit from treatment with lactitol and lactulose in terms of their psychometric performance. The feasibility and benefits of long-term treatment for this condition need to be elucidated.