Background: In this prespecified AIDA-trial sub-study we investigate the clinical performance of absorb bioresorbable vascular scaffold (BVS) compared to Xience everolimus-eluting stent (EES) in routine percutaneous coronary intervention (PCI) in patients with diabetes mellitus (DM) at complete 3-year follow-up. Methods and results: All 1,845 randomized patients were subdivided by medical history with DM or without DM. Of the 924 Absorb BVS patients, 171 (18.5%) patients had DM, of which 65 (38.0%) were treated with insulin (iTDM). Of the 921 Xience EES patients, 153 (16.6%) patients had DM, of which 45 (29.4%) were insulin-treated diabetes mellitus (iTDM). Target vessel failure (TVF), composite of cardiac death, target vessel myocardial infarction, and target vessel revascularization, occurred in 18.7% of diabetic patients treated with Absorb patients versus in 18.0% patients treated with Xience EES (p =.840). In nondiabetics the rates of TVF were 12.3% in Absorb BVS versus 11.0% in Xience EES (p =.391). Definite/probable device thrombosis occurred more frequently in Absorb BVS compared to Xience EES in both diabetic and nondiabetic patients (4.8% versus 0.7%; p =.028 and 3.2% vs. 0.5%; p <.001, respectively). Conclusions: In routine PCI practice, both Absorb BVS and Xience EES have worse clinical outcomes in diabetic patients as compared to nondiabetic patients. Throughout all clinical presentations, Absorb BVS was associated with higher rates of device thrombosis at 3-year follow-up.