Few studies directly compared drug-eluting stents and bare-metal stents (BMSs) in diabetic patients. DESSERT was an Italian multicenter randomized trial to show the efficacy of sirolimus-eluting stents (SESs) compared with BMSs in de novo lesions of diabetic patients treated with insulin and/or oral antidiabetics for > or =3 months on top of glycoprotein IIb/IIIa inhibitors. The primary end point was in-stent late lumen loss, assessed using centralized quantitative coronary angiography at 8-month follow-up. Centrally adjudicated composite major adverse cardiac events (MACEs) and target-vessel failure (TVF; death, treated vessel-related acute myocardial infarction, and target-vessel revascularization) at 30 days and 9 and 12 months were secondary end points. Seventy-five patients were randomly assigned to an SES (109 lesions), and 75 (109 lesions), to a BMS. The 2 groups were well balanced for clinical, anatomic, and procedural characteristics. In-stent late lumen loss decreased from 0.96 +/- 0.61 mm for BMSs to 0.14 +/- 0.33 for SESs (p <0.001), and in-segment binary restenosis was 38.8% versus 3.6%, respectively (p <0.001). Twelve-month clinical events were significantly lower in the sirolimus group: MACEs 22.1% versus 40% (p = 0.023), target-lesion revascularization 5.9% versus 30% (p <0.001), and TVF 14.7% versus 34.3% (p = 0.008). At multivariate analysis, stent type was confirmed as an independent predictor of in-segment late loss (p <0.001), binary restenosis (p <0.001), 12-month TVF (p = 0.010), and 12-month MACEs (p = 0.037). In conclusion, the randomized DESSERT showed SESs to be safe and effective in decreasing both angiographic parameters of restenosis and incidence of MACEs compared with BMSs in diabetic patients with de novo 1- or 2-vessel coronary stenoses.