Background: The ESPAC-3 trial showed that adjuvant gemcitabine is the standard of care based on similar survival to and less toxicity than adjuvant 5-fluorouracil/folinic acid in patients with resected pancreatic cancer. Other clinical trials have shown better survival and tumour response with gemcitabine and capecitabine than with gemcitabine alone in advanced or metastatic pancreatic cancer. We aimed to determine the efficacy and safety of gemcitabine and capecitabine compared with gemcitabine monotherapy for resected pancreatic cancer. Methods: We did a phase 3, two-group, open-label, multicentre, randomised clinical trial at 92 hospitals in England, Scotland, Wales, Germany, France, and Sweden. Eligible patients were aged 18 years or older and had undergone complete macroscopic resection for ductal adenocarcinoma of the pancreas (R0 or R1 resection). We randomly assigned patients (1:1) within 12 weeks of surgery to receive six cycles of either 1000 mg/m2 gemcitabine alone administered once a week for three of every 4 weeks (one cycle) or with 1660 mg/m2 oral capecitabine administered for 21 days followed by 7 days' rest (one cycle). Randomisation was based on a minimisation routine, and country was used as a stratification factor. The primary endpoint was overall survival, measured as the time from randomisation until death from any cause, and assessed in the intention-to-treat population. Toxicity was analysed in all patients who received trial treatment. This trial was registered with the EudraCT, number 2007-004299-38, and ISRCTN, number ISRCTN96397434. Findings: Of 732 patients enrolled, 730 were included in the final analysis. Of these, 366 were randomly assigned to receive gemcitabine and 364 to gemcitabine plus capecitabine. The Independent Data and Safety Monitoring Committee requested reporting of the results after there were 458 (95%) of a target of 480 deaths. The median overall survival for patients in the gemcitabine plus capecitabine group was 28·0 months (95% CI 23·5-31·5) compared with 25·5 months (22·7-27·9) in the gemcitabine group (hazard ratio 0·82 [95% CI 0·68-0·98], p=0·032). 608 grade 3-4 adverse events were reported by 226 of 359 patients in the gemcitabine plus capecitabine group compared with 481 grade 3-4 adverse events in 196 of 366 patients in the gemcitabine group. Interpretation: The adjuvant combination of gemcitabine and capecitabine should be the new standard of care following resection for pancreatic ductal adenocarcinoma. Funding: Cancer Research UK.