Background: Several studies have investigated the diagnostic performance of computed tomography coronary angiography (CTCA) for the detection of significant coronary artery disease (CAD). These studies were performed in patients that were already referred for invasive coronary angiography (ICA) and prevalence of significant CAD was high. Although the negative predictive value of CTCA was consistently high, a wide range of positive predictive values (PPVs) was reported. Thus, the PPV of CTCA in patients that undergo CTCA as part of a clinical diagnostic evaluation remains unclear. This study investigated the PPV of CTCA for the detection of significant CAD in clinical practice. Methods: A total of 181 patients with low to intermediate pre-test probability CAD that were referred for non-invasive evaluation of chest pain underwent 64-slice CTCA. CTCA was scored per segment as normal, non-obstructive CAD or obstructive CAD (> 50% diameter stenosis). All patients with obstructive CAD according to CTCA, underwent ICA. Significant CAD was defined as > 50% diameter stenosis on ICA. Results: According to CTCA, 65 (35.9%) patients had obstructive CAD. In 26 (14.4%) patients, significant CAD was found by ICA. The PPV for detection of significant CAD per patient, per vessel and per segment were 40.0% (26/65, 95% CI: 30.6-50.2%), 31.3% (36/115, 95% CI: 24.7-38.8%) and 25.5% (42/165; 95% CI: 20.3-31.4%), respectively. Conclusions: The PPV of CTCA for detection of significant CAD in patients with low to intermediate probability CAD that are clinically referred for non-invasive evaluation of chest pain is markedly lower than generally reported. © 2010 Elsevier Ireland Ltd. All rights reserved.