Liver transplantation is proposed as the best therapy for early hepatocellular carcinoma in cirrhotic patients. However, the confrontation with the results obtained by surgical resection has never been done on an intention-to-treat basis. Between 1989 and 1997, 164 out of 1,265 patients with hepatocellular carcinoma were evaluated for surgery. Seventy-seven (48 men, mean 61 years of age, 74 Child-Pugh class A, size 33 +/- 18 mm) were resected (first line option) and 87 (65 men, mean 55 years of age, 50 Child-Pugh class B/C, size 24 +/- 14 mm) were selected for transplantation. The 1-, 3-, and 5-year "intention-to-treat" survival was 85%, 62%, and 51% for resection and 84%, 69%, and 69% for transplantation (8 drop-outs on waiting list). Bilirubin and clinically relevant portal hypertension were independent survival predictors after resection. Thereby, the 5-year survival of the best candidates (absence of clinically relevant portal hypertension, n = 35) was 74%, whereas it was 25% for the worst candidates (portal hypertension and bilirubin >/=1 mg/dL, n = 27) (P <.00001). The variable "drop-out on waiting list" was the sole survival predictor after transplantation. The 2-year survival rate of patients evaluated for transplantation was 84% in the 1989 to 1995 period (mean waiting time, 62 days; no drop-outs) and 54% during 1996 to 1997 (mean waiting time, 162 days; 8 drop-outs)(P <.003). This outcome was significantly lower than that of the best candidates for resection (P =.002). In conclusion, a proper selection of candidates for resection promotes better results than transplantation, in which the results are significantly hampered by the growing incidence of drop-outs because of the increasing waiting time.