Cholangiocarcinoma presents many challenges. Prognosis is thought to be determined by conventional predictors of survival; margin status, pathologic criteria, stage, and comorbid disease. Ninety-four patients, 57 males and 37 females, underwent resections for cholangiocarcinoma between 1989 and 2000. Thirty-two patients (34%) had distal tumors, 10 had midduct lesions, and 52 had proximal/intrahepatic lesions. Thirty-four patients underwent pancreaticoduodenectomies, 23 bile duct resections alone, and 37 bile duct and concomitant hepatic resections. Tumor location did not influence mean survival (distal, 28 months +/- 23; midduct, 28 months +/- 21; and proximal, 31 months +/- 36). Operation undertaken did not alter survival (bile duct resection, 30 months +/- 37; pancreaticoduodenectomy, 27 months +/- 23; and concomitant bile duct/hepatic resection, 32 months +/- 32). TNM stage failed to predict survival: 5 stage I (29 months +/- 22), 12 stage II (41 months +/- 33), 12 stage III (33 months +/- 19), and 64 stage IV (27 months +/- 32). Tumor size did not influence survival: T1-2 (32 months +/- 33) versus T3-4 lesions (29 months +/- 25). Mean survival with negative margin (n = 67) was 34 months +/- 33, whereas microscopically positive (n = 13, 23.9 months +/- 25) or grossly positive (n = 14, 20.4 months +/- 20) margins were predictive of significantly shorter survival (P < 0.03). Adjuvant treatment (n = 41) was associated with significantly longer survival (40.5 months +/- 36) than those who received no further therapy (n = 53; 24 months +/- 24) (P = 0.05). TNM stage, tumor size, operation undertaken, and location were not associated with duration of survival after resection. Margin status was associated with duration of survival, though extended survival is possible even with positive margins. Advanced stage should not preclude aggressive resection. Without specific contraindications, an aggressive operative approach is advocated followed by adjuvant therapy.