We conducted a lesion-based retrospective sub-analyses of diabetes mellitus (DM), diffuse long lesions (stented segment ≥40 mm; LLs), and small vessels (SVs; reference diameter ≤2.6 mm) in patients who received sirolimus- (SESs) or paclitaxel-eluting stents (PESs) for nonrandom treatment of de novo native coronary stenosis in a clinical practice setting. During the period from May 2007 to February 2009, 490 of 682 PES-treated and 293 of 386 SES-treated lesions were angiographically followed up within 1500 days of PCI, and the retrospective investigation was conducted in April 2013. The frequencies of target lesion revascularization (TLR; any recurrent PCI including both marginal stent restenosis) and binary in-stent restenosis (percentage diameter of in-stent stenosis >50 %) upon follow-up angiography, evaluated by adjusting 25 baseline variables using propensity score matching analysis, after placement of SESs and PESs were the following: DM (n = 124 per arm), 14.5 vs. 15.3 % (p = 0.842), and 14.5 vs. 16.1 % (0.856); LLs (n = 81), 16.0 vs. 21.0 % (0.433), and 12.3 vs. 22.2 % (0.117); SVs (n = 107), 11.2 vs. 29.9 % (<0.001), and 11.2 vs. 30.8 % (<0.001), respectively. The p values of log-rank tests for the cumulative TLR-free ratios after SES and PES placement were 0.504 in DM, 0.625 in LLs, and <0.001 in SVs group, respectively. Thus, compared to PES, SES showed the equivalent efficacy for DM, the tendency to be superior for LLs due to approximately 24–45 % reductions in TLR and binary restenosis rates, and the promising superiority for SVs on the angiographic outcomes during a long-term observational interval.