Our uncontrolled phase II study was aimed at assessing the efficacy of transarterial embolization in patients with hepatocellular carcinoma and to determine the parameters associated with a favorable response to treatment, improved survival or both. Fifty consecutive patients (25 corresponding to Okuda's stage I and 25 to stage II) with hepatocellular carcinoma (41 being multinodular or massive) were included. Transarterial embolization induced a self-limited postembolization syndrome that was well tolerated. Nevertheless, three patients died shortly after the procedure because of tumor progression (two cases) or progressive liver failure. A favorable response (extensive necrosis with reduction of tumor area greater than 50%) was achieved in 81% of the cases, and this result was independently (p < 0.05) related to a preserved performance status and to a lower alphafetoprotein concentration. The survival of the patients at 1 and 2 yr was 65% and 38%, respectively, better than the expected survival according to a mathematical model obtained from a historical series of untreated cases (42% and 20%, respectively). Cox regression analysis disclosed that both a favorable therapeutic response and a preserved physical condition (reflected by performance status of 0 or 1) were independently associated with better survival (regression coefficient -2.248 and 0.869, respectively). These data indicate that transarterial embolization has a marked antitumoral effect in patients with inoperable hepatocellular carcinoma and that the therapeutic success is associated with improved survival. Nevertheless, because the potential benefit for survival observed in this uncontrolled study appears to be moderate, prospective controlled trials to ascertain the real usefulness of this therapeutic approach are mandatory. (HEPATOLOGY 1994;20:643-650).