Background: Although some retrospective studies have suggested the value of adjuvant therapy, no recommended standard exists in bile duct cancer. The aim of this study was to test the hypothesis that adjuvant gemcitabine chemotherapy would improve survival probability in resected bile duct cancer. Methods: This was a randomized phase III trial. Patients with resected bile duct cancer were assigned randomly to gemcitabine and observation groups, which were balanced with respect to lymph node status, residual tumour status and tumour location. Gemcitabine was given intravenously at a dose of 1000 mg/m2, administered on days 1, 8 and 15 every 4 weeks for six cycles. The primary endpoint was overall survival, and secondary endpoints were relapse-free survival, subgroup analysis and toxicity. Results: Some 225 patients were included (117 gemcitabine, 108 observation). Baseline characteristics were well balanced between the gemcitabine and observation groups. There were no significant differences in overall survival (median 62·3 versus 63·8 months respectively; hazard ratio 1·01, 95 per cent c.i. 0·70 to 1·45; P = 0·964) and relapse-free survival (median 36·0 versus 39·9 months; hazard ratio 0·93, 0·66 to 1·32; P = 0·693). There were no survival differences between the two groups in subsets stratified by lymph node status and margin status. Although haematological toxicity occurred frequently in the gemcitabine group, most toxicities were transient, and grade 3/4 non-haematological toxicity was rare. Conclusion: The survival probability in patients with resected bile duct cancer was not significantly different between the gemcitabine adjuvant chemotherapy group and the observation group. Registration number: UMIN 000000820 (http://www.umin.ac.jp/).