Summary Background The best way to manage restenosis in patients who have previously received a drug-eluting stent is unknown. We investigated the efficacy of paclitaxel-eluting balloons (PEB), paclitaxel-eluting stents (PES), and balloon angioplasty in these patients. Methods In this randomised, open-label trial, we enrolled patients older than 18 years with restenosis of at least 50% after implantation of any limus-eluting stent at three centres in Germany between Aug 3, 2009, and Oct 27, 2011. Patients were randomly assigned (1:1:1; stratified according to centre) to receive PEB, PES, or balloon angioplasty alone by means of sealed, opaque envelopes containing a computer-generated sequence. Patients and investigators were not masked to treatment allocation, but events and angiograms were assessed by individuals who were masked. The primary endpoint was diameter stenosis at follow-up angiography at 6–8 months. Primary analysis was done by intention to treat. This trial is registered with ClinicalTrials.gov , number NCT00987324. Findings We enrolled 402 patients, of whom 137 (34%) were assigned to PEB, 131 (33%) to PES, and 134 (33%) to balloon angioplasty. Follow-up angiography at 6-8 months was available for 338 (84%) patients. PEB was non-inferior to PES in terms of diameter stenosis (38·0% SD 21·5 vs 37·4% 21·8; difference 0·6%, one-sided 95% CI 4·9%; pnon-inferiority =0·007; non-inferiority margin of 7%). Findings were consistent in per-protocol analysis (pnon-inferiority =0·011). PEB and PES were superior to balloon angioplasty alone (54·1% 25·0; psuperiority <0·0001 for both comparisons). Frequency of death, myocardial infarction, or target lesion thrombosis did not differ between groups. Interpretation By obviating the need for additional stent implantation, PEB could be a useful treatment for patients with restenosis after implantation of a drug-eluting stent. Funding Deutsches Herzzentrum.