The results of surgery on Klatskin tumors have improved during the era of the combined liver resection. However, some surgeons still have a negative point of view on liver resection because of the locally infiltrative characteristics of cancer and the high mortality and morbidity. We treated 98 patients with a Klatskin tumor between 1990 and 1996. The mean age was 57 year, and the sex ratio (M : F) was 2.2 : 1. Among them, 27 patients (27.6%) received a combined liver resection, 11 patients (11.2%) received a segmental resection of the extrahepatic bile duct, and the other 60 patients (61.2%) had non resective procedures (a bypass operation, an intubational operation, or a percutaneous drainage procedure). The mean survival of the combined liver resection group was 30.0 months-higher than those of the other groups (17.1 months for the segmental-resection group, 14.0 months for the non resective procedures group)(p<0.05). The cumulative 1-, 2-, and 3-year survival rates in patients undergoing a combined liver resection were 95.5%, 83.9%, and 55.9% respectively (significantly higher than the 72.7%, 11.4%, and 0% in the patients undergoing a resection and the 41.1%, 11.4%, and 5.7% in the patients undergoing a non resective procedures). There was no difference in the survival rates between the segmental-resection and non resective-procedure group. In the segmental-resection group, 5 patients had a negative resection margin, and 6 patients had a positive resection margin. The mean survivals were 18.4 and 16.0 months, respectively (p>0.05). The morbidity rate for the combined liver resection was 74%--higher than that for the segmental resection (45%) and for the non resective-procedure group (30%)(p<0.05). There was one operative mortality (1.0%), the patient died due to postoperative sepsis after a U-tube intubational operation. Based on a univariate analysis, significant prognostic factors after liver resection were lymph node involvement, and bilirubin (>5 mg/dl) at admission. A multivariate analysis showed no prognostic significance for either lymph node involvement or bilirubin at admission. These results indicate that segmental resection is not a curable treatment modality and that only aggressive radical surgery, including a hepatectomy, gives any chance of a cure for a Klatskin tumor. We conclude that combined radical liver resection is the treatment of choice for a Klatskin tumor and that accurate preoperative diagnosis and cautious perioperative care decrease the mortality and morbidity of a hepatectomy.