Purpose: To estimate the sensitivity and specificity of a diagnostic test if multiple measurements within individual patients are performed using a bivariate meta-analysis, which takes into account correlated measurements and heterogeneity across patients. Methods: Fifty patients underwent multi-detector CT angiography (MDCTA) and invasive coronary angiography, the reference test, for the assessment of coronary artery disease. The coronary artery tree was subdivided into 17 segments per patient. Uninterpretable segments were excluded. The sensitivity and specificity of MDCTA were calculated on a segment basis (n=703), assuming independent measurements. Next, the sensitivity and specificity were calculated per patient and pooled using a random-effects bivariate meta-analysis. This meta-analytical method adjusts for the possible correlation between multiple measurements within one patient and accounts for heterogeneity across patients. In this bivariate approach, the relationship between sensitivity and specificity is also taken into account. The within patient variability is modeled using a binomial distribution. The sensitivity and specificity estimates and 95% confidence intervals of the segment analysis and the bivariate meta-analysis were compared. Results: Sensitivity was 53.8% (95%CI 43.1-64.2%) using the segment based analysis and 52.3% (95%CI 35.9-68.2%) using the bivariate approach. For specificity the results were 88.3% (95%CI 85.5-90.8%) and 89.0% (95%CI 85.4-91.8) respectively. These results suggest that the 95% confidence interval is underestimated in the segment-based analysis because of correlation between measurements. Conclusion: If multiple measurements of a diagnostic test are performed per patient, sensitivity and specificity can be estimated using a bivariate meta-analytical method which adjusts for possible correlation between measurements.