Intravascular ultrasound (IVUS) imaging has been used to optimize stent implantation in coronary arteries, but the criteria used were chosen on an empiric basis. The aim of this study was to determine whether any of these criteria have an independent role in predicting the probability of freedom from restenosis. The study population consisted of 425 patients (496 lesions) who underwent angiographically successful IVUS-guided stenting. Five IVUS criteria were studied: (1) intrastent minimal lumen cross-sectional area (ISMLCSA) > or =9 mm2; (2) ISMLCSA (> or =9 mm2 and > or =80% of average reference lumen cross-sectional area [CSA]); (3) ISMLCSA > or =90% of average reference lumen CSA; (4) ISMLCSA > or =90% of distal reference lumen CSA; and (5) ISMLCSA > or =55% of average reference vessel CSA. These criteria were met in 33%, 29%, 68%, 82%, and 69% of lesions, respectively. Angiographic follow-up was performed in 335 of 421 eligible patients (80%) at 5.3 +/- 2.7 months. An absolute ISMLCSA > or =9 mm2 was associated with the lowest restenosis, but this criterion was primarily achieved in large vessels. The only criterion that was associated with higher probability of freedom from restenosis independently from vessel size was an ISMLCSA > or =55% of average reference vessel CSA. Therefore, when IVUS is used to guide stent implantation an effort should be made to achieve the largest lumen safely possible. An ISMLCSA > or =55% of the average reference vessel CSA seems to be the most appropriate criterion in terms of frequency of achievement and in terms of increasing the probability of freedom from restenosis.