Multislice CT coronary angiography (CT-CA) has emerged as a potential imaging method for coronary artery disease. This study aimed to ascertain the accuracy of 16-slice CT in the diagnosis of significant coronary stenosis (≥50% reduction of lumen diameter). This mixed retrospective/prospective observational study compared 95 paired 16-slice CT-CA and fluoroscopic coronary angiography (FCA) sets. A cardiologist and a radiologist blinded to the FCA findings evaluated CT-CA images independently by visual estimation. Disagreement between these reporters was arbitrated by a third CT reporter (a cardiologist). A separate cardiologist blinded to CT-CA findings assessed FCA by visual estimation. Of 1161 coronary segments assessable on FCA, 1103 segments (95%) were assessable on CT-CA. The CT-CA correctly diagnosed 147/180 segments with significant stenoses (sensitivity = 82%) and correctly identified 874/923 coronary segments without significant stenoses (specificity = 95%). The positive and negative predictive values of CT-CA in the diagnosis of coronary segment with significant stenosis were 75 and 96%, respectively. On patient-based analysis, CT-CA correctly identified all 68 studies with at least one vessel with significant stenosis (sensitivity = 100%; specificity = 83%). The positive and negative predictive values of CT-CA in identifying patients with significant coronary stenosis were 94 and 100%, respectively. The 16-slice CT-CA showed moderately good sensitivity but very high specificity and negative predictive value in the diagnosis of significant coronary stenosis. The CT-CA would appear to be a useful ‘rule-out’ test for patients with low-risk profile for ischaemic heart disease.