Selection of coronary stent length varies from covering only the zone of maximum obstruction to stenting from normal- to normal-appearing vessels. With bare metal stenting, for any given lesion there is a high restenotic risk associated with longer stent length. With drug-eluting stents, the relation between stent length and restenosis has not been evaluated. In the angiographic follow-up cohort of the SIRIUS trial that compared the sirolimus-eluting Bx Velocity stent with the standard Bx Velocity stent (n = 699), we constructed a multiple regression model to predict 8-month percent diameter stenosis using the main effects of lesion length and excess stent length beyond the lesion length and adjusting for known predictors of restenosis. Stent length exceeded lesion length in 94% of lesions overall. Mean difference in length was 8.3 +/- 8.3 mm (mean lesion length 14.6 +/- 5.9 mm, mean stent length 22.9 +/- 9.6 mm). Stented lesion length and excess stent length were associated with absolute increases in percent diameter stenosis per 10 mm of 9.1% (p <0.0001) and 3.6% (p = 0.053) in the bare metal arm and 3.5% (p = 0.047) and 2.1% (p = 0.040) in the sirolimus-eluting stent arm. Although the effects of lesion length and excess stent length on restenosis were markedly decreased with sirolimus-eluting stents (vs bare metal stents), a small restenotic penalty is still paid for excessive stent length. Longer stent-to-lesion length strategies should be used only when a shorter stent is likely to result in incomplete lesion coverage and edge dissection, a strong determinant of stent thrombosis.