Summary Background In head-to-head comparisons of coronary drug-eluting stents, the primary endpoint is traditionally assessed after 9–12 months. However, the optimum timepoint for this assessment remains unclear. In this study, we assessed clinical outcomes at up to 5 years' follow-up in patients who received two different types of drug-eluting stents. Methods We undertook this multicentre, open-label, randomised superiority trial at five percutaneous coronary intervention centres in Denmark. We randomly allocated 2332 eligible adult patients (≥18 years of age) with an indication for drug-eluting stent implantation to the zotarolimus-eluting Endeavor Sprint stent (Medtronic, Santa Rosa, CA, USA) or the sirolimus-eluting Cypher Select Plus stent (Cordis, Johnson & Johnson, Warren, NJ, USA). Randomisation of participants was achieved by computer-generated block randomisation and a telephone allocation service. The primary endpoint of the SORT OUT III study was a composite of major adverse cardiac events—cardiac death, myocardial infarction, and target vessel revascularisation—at 9 months' follow-up. In this study, endpoints included the occurrence of major adverse cardiac events and definite stent thrombosis at follow-up times of up to 5 years. Analysis was by intention to treat. The trial is registered with ClinicalTrials.gov , number NCT00660478. Findings We randomly allocated 1162 patients to receive the zotarolimus-eluting stent and 1170 to the sirolimus-eluting stent. At 5-year follow-up, rates of major adverse cardiac events were similar in patients treated with both types of stents (zotarolimus-eluting stents 197/1162 17·0% vs sirolimus-eluting stents 182/1170 15·6%; odds ratio OR 1·10, 95% CI 0·88–1·37; p=0·40). This finding was indicative of the directly contrasting results for rates of major adverse cardiac events at 1-year follow up (zotarolimus 93/1162 8·0% vs sirolimus 46/1170 3·9%; OR 2·13, 95% CI 1·48–3·07; p<0·0001) compared with those at follow-up between 1 and 5 years (104 9·0% vs 136 11·6%; OR 0·78, 95% CI 0·59–1·02; p=0·071). At 1-year follow-up, definite stent thrombosis was more frequent after implantation of the zotarolimus-eluting stent (13/1162 1·1%) than the sirolimus-eluting stent (4/1170 0·3%; OR 3·34, 95% CI 1·08–10·3; p=0·036), whereas the opposite finding was recorded for between 1 and 5 years' follow-up (zotarolimus-eluting stent 1/1162 0·1% vs sirolimus-eluting stent 21/1170 1·8%, OR 0·05, 95% CI 0·01–0·36; p=0·003). 26 of 88 (30%) target lesion revascularisations in the zotarolimus-eluting stent group occurred between 1 and 5 years' follow-up, whereas 54 of 70 (77%) of those in the sirolimus-eluting stent group occurred during this follow-up period. Interpretation The superiority of sirolimus-eluting stents compared with zotarolimus-eluting stents at 1-year follow-up was lost after 5 years. The traditional 1-year primary endpoint assessment therefore might be insufficient to predict 5-year clinical outcomes in patients treated with coronary drug-eluting stent implantation. Funding Cordis and Medtronic.