In a retrospective study the survival of 28 patients with hepatocellular carcinoma, 25 of them with underlying cirrhosis, submitted to surgical resection was compared with the survival of 28 untreated patients, matched for variables known to bear independent prognostic value and therefore sharing the same baseline prognosis. Diagnosis was made in the same time period for both groups of patients. In addition, to further evaluate the effects of tumor resection on survival, the outcome of operated patients was also compared to their expected survival. This was derived from a mathematical model which takes into account the regression coefficients of the variables previously shown to be independently related to the survival of untreated patients with hepatocellular carcinoma. The median survival for resected patients was 27.1 months, which was significantly better than untreated controls (12.4 months; p less than 0.003). Median survival for patients submitted to resection and with tumors smaller than 5 cm was 35.8 months, while the median survival for untreated cases was 14.6 months p less than 0.0005. The comparison of observed survival (82% at one year and 73% at two years) and statistically expected survival (58% and 34%, respectively) further indicated that surgical resection effectively improves prognosis in Western patients with hepatocellular carcinoma. Thus, early detection of small tumors in the population at risk appears to be justified.