Late luminal loss after percutaneous coronary intervention, although not normally distributed, has been shown to be monotonically correlated with the probability of binary angiographic restenosis after drug-eluting stent implantation. A recently proposed method has been shown to predict the restenosis rate after sirolimus-eluting stent and paclitaxel-eluting stent implantation using a power transformation of the value of late loss obtained after implantation of the 2 stent types. We used the same method to compute and compare restenosis rates from late loss values observed in the “head-to-head” randomized sirolimus-eluting stent versus paclitaxel-eluting stent comparisons available thus far. Our results showed that the model proposed has a poor overall ability to predict the real incidence and relative risk of restenosis because the use of late loss as an end point tended to overestimate the difference in restenosis, and the derived effect estimate of 1 stent compared with that of the other seemed to be overemphasized with respect to the real risk. We thus believe that further investigations are needed to appraise the real clinical effect of late loss and its reliability as an end point in direct comparative drug-eluting stent trials before its use can be recommended as a clinical surrogate.