The diagnostic performance of in-stent restenosis (ISR) by 64-slice multidetector CT coronary angiography (CTCA) has been reported to be influenced by multiple factors. We evaluated individual factors (stent diameter, material and strut thickness) and therefore determined the proper population for follow-up by using this modality. A total of 171 stents were evaluated in 83 consecutive patients with stents imaged with CTCA and conventional coronary angiography. The stent diameter ranged from 2.25 mm to 4.5 mm. 2 models of stainless steel (Taxus Liberte (Boston Scientific, US), 56 stents and Cypher Select (Cordis, US), 34 stents) and 2 models of cobalt alloy (Endeavor (Medtronic, US), 33 stents and Firebird2 (MicroPort, China), 48 stents) were included. By comparing to conventional coronary angiography, the image quality and diagnostic accuracy for ISR were evaluated. The image quality of Taxus, Endeavor and Firebird are markedly better than Cypher in large caliber group (≧3.0 mm) (P < 0.001). Except for Cypher, all other stents with diameter ≧3.0 mm showed excellent diagnostic accuracy (sensitivity 100%, specificity 94.4-96% whereas stents with diameter <3.0 mm had poor diagnostic accuracy (sensitivity 100%, specificity 33.3-70%). Cypher is the stent with thickest strut in our study, and showed reduced image quality and diagnostic accuracy in all stent size, due to large number of unassessable stents. Among 16 binary ISR, 12 lesions were correctly diagnosed by CTCA while the other 4 lesions were unassessable. The main reason for low specificity in small caliber group is the large number of unassessable stents. CTCA has high diagnostic accuracy to identify ISR in selected stents with a diameter of ≧3.0 mm.