Acute liver failure (ALF) and hepatic encephalopathy (HE) can lead to an elevated intracranial pressure (ICP) and death within days. The impaired liver function increases the risks of invasive ICP monitoring, whereas noninvasive methods remain inadequate. The purpose of our study was to explore reliable noninvasive methods of neuromonitoring for patients with ALF in the intensive care unit (ICU) setting; more specifically, we wanted to track changes in HE and predict the outcomes of ALF patients treated with albumin dialysis. The study included 20 patients with severe ALF at admission who had been referred to the ICU of the liver transplantation (LT) center for albumin dialysis treatment and evaluation for transplantation. Data were collected from all study patients in the form of continuous frontal electroencephalography (EEG) recordings and transcranial Doppler (TCD) measurements of cerebral blood flow. Among the studied EEG variables, the 50% spectral edge frequency decreased and the delta power increased as the HE stage increased. Both variables were predictive of the stage of HE [prediction probability (P K ) of 50% spectral edge frequency 5 0.23, standard error (SE) 5 0.03; P K of delta power 5 0.76, SE 5 0.03]. The total wavelet subband entropy, a novel variable that we used for tracking abnormal EEG activity, predicted the outcome of ALF patients treated with albumin dialysis (P K 5 0.88, SE 5 0.09). With a threshold value of 1.6, the TCD pulsatility index had an odds ratio of 1.1 (95% confidence interval 5 0.1-9.3) for a poor outcome (LT or death). In conclusion, EEG variables are useful for the monitoring of HE and can be used to predict outcomes of ALF. TCD measurements do not predict patient outcomes.