The objective of this pilot controlled study was to evaluate the extracorporeal liver assist device (ELAD) in patients with acute liver failure who were judged to still have a significant chance of survival (approximately 50%) and in those who had already fulfilled criteria for transplantation. Twenty-four patients were divided into two groups, 17 with a potentially recoverable lesion (group I) and 7 listed for transplantation (group II), and then randomly allocated to ELAD haemoperfusion or control. The median period of ELAD haemoperfusion was 72 hours (range 3-168 h). Biocompatibility of the device was good, with no acceleration in platelet consumption, and haemodynamic stability was maintained. Two patients were withdrawn from the study because of worsening of preexisting disseminated intravascular coagulation in one case and a hypersensitivity reaction in the other. Deterioration with respect to encephalopathy grade was more frequent in the control patients, 7 of 12 (58%), than in the ELAD-treated patients, 3 of 12 (25%). In group I where survival for the ELAD cases was 7 of 9 (78%), there was a higher than expected survival in the controls, 6 of 8 (75%). For group II cases, survival was 1 of 3 (33%) for the ELAD-treated patients, and 1 of 4 (25%) for the controls. Both of the survivors underwent transplantation. Assessment of additive function for the device revealed an improvement in galactose elimination capacity after 6 hours of haemoperfusion. Based on the results of this pilot-controlled trial, better indices of prognosis will be required, in addition to those used to select for transplantation, if patients at an earlier stage of clinical deterioration are to be included in future studies.