We performed a prospective study with ultrasound (US) follow-up on a population at risk for hepatocellular carcinoma (HCC) to evaluate the possibilities of diagnosis of HCC at an early stage. We studied 360 patients; 254 of those patients had cirrhosis of the liver (of which 167 had HCV, 28 had HBV, 53 were alcoholic, and six had autoimmune disease), and the remaining 106 patients had chronic hepatitis. US scan was performed and alpha-fetoprotein was measured every 6 months. Mean duration of follow-up was 56 months (range 18-72). We compared the results of HCC detection with those of 2170 patients with cirrhosis or chronic hepatitis observed with US in the same period, outside the follow-up protocol. In 24 of the follow-up patients, hepatic lesions were detected with US and proved to be HCC with fine needle biopsy, with a cumulative incidence of 6.6% and an annual incidence of 1.4%. The HCC was unifocal in 18 patients, always with diameter < or = 3 cm (small HCC). All but one of the 24 patients who developed HCC had liver cirrhosis, and the remaining patient had HBV-correlated chronic hepatitis. In the group of 2170 patients without serial follow-up, the percentage of detected unifocal tumors with diameters < or = 3 cm was only 15.5%, compared with 75% in the patients with serial follow-up. The US follow-up of a population at risk for HCC, consisting of patients with chronic liver diseases, made it possible to diagnose tumor < or = 3 cm in a high percentage of cases, with a statistically highly significant difference (chi 2, p = 0.000) compared with the patients not in the follow-up. Follow-up could increase the percentage of HCCs detected at a potentially curable stage. Cirrhosis is confirmed as being the main risk cause, regardless of etiology.