In comparing the restenosis rates among different interventions, 1 potential confounder might be the differences in the vessels treated, as dictated by the technical limitations of particular devices. The purpose of this study was to use current "acute gain-late loss" analysis to examine what influence vessel selection has on the restenosis rates seen after coronary stenting or directional atherectomy. The minimal luminal diameter of native coronary lesions was measured before and immediately after intervention in 102 single Palmaz-Schatz stents and 347 atherectomies, 367 (82%) of which had repeat angiographic measurement 6 months after intervention. Atherectomy-treated lesions had a higher proportion of left anterior descending to right coronary arteries (68 vs 24%) compared with stents (31 vs 54%), p < 0.001. Although subsequent restenosis rates were similar for stenting (25%) and atherectomy (30%, p = 0.42), left anterior descending versus right coronary lesions had a significantly higher restenosis rate for the overall group (35 vs 18%, p = 0.009), for stents (44 vs 13%, p = 0.008) and for atherectomy (35 vs 22%, p = 0.10), respectively. Multivariable analysis demonstrated that postprocedure luminal diameter (p = 0.03, p = 0.009) and coronary location (the proportion of left anterior descending vessels treated, p = 0.002, p < 0.001), but not device type (stent vs atherectomy), were strong independent determinants of restenosis according to both binary (> 50% diameter stenosis) and continuous (late percent stenosis) definitions.(ABSTRACT TRUNCATED AT 250 WORDS)