A prospective study of hepatic encephalopathy (HE) including neuropsychiatric and psychometric evaluation, electroencephalography, and determination of arterial ammonia levels was performed in 55 cirrhotic patients treated consecutively by transjugular intrahepatic portosystemic shunt (TIPS). The cumulative HE rate increased from 23.6% within the 3-month interval before TIPS to 50. 9% within the first 3-month interval post-TIPS (P = .003). Significant and independent predictors of HE post-TIPS were the presence of HE pre-TIPS and reduced liver function. The cumulative HE rate declined in the second 3-month interval post-TIPS and reached the pre-TIPS level. Chronic forms of HE exceeding grade I were not observed. In a subgroup of 22 nonencephalopathic TIPS patients, the prevalence of subclinical HE did not change after TIPS. Among individual psychometric tests, the block design test gave the highest proportion of pathological results (about 50%), whereas selective reminding gave the lowest (10%-25%). Electroencephalography (EEG) showed a temporary increase of pathological results at 1 month after TIPS, when patients with overt HE (grade I) were included (proportion of 21.1% before vs. 57.1%, P = .005). Arterial ammonia concentration increased from a mean of 94 +/- 26 microgram/dL to 140 +/- 28 microgram/dL at 3 months after TIPS (P < .001). Elevated ammonia levels persisted. TIPS led to a temporary increase of HE incidence within 3 months. The decline of the HE rate beyond 3 months despite a sustained increase of arterial ammonia levels could not entirely be explained by reduction of shunt flow, nor by alteration of liver function. Instead, cerebral adaptation to gut-derived neurotoxins might be anticipated.