Non-invasive evaluation of coronary artery disease (CAD) in patients with left bundle branch block (LBBB) has limitations inherent to different tests, and the relative merits of these tests are unclear. This meta-analysis assessed the accuracy of the frequently used non-invasive techniques, including exercise electrocardiography (ECG), myocardial perfusion imaging (MPI) and stress echocardiography (SE), for detection of CAD and prediction of cardiac events in patients with LBBB. A review was conducted of all reports on detection of CAD and prediction of cardiac events in patients with LBBB (published between January 1970 and December 2004), and revealed 55 diagnostic and nine prognostic reports with sufficient details to calculate test accuracy. Weighted (by sample size) sensitivity and specificity were calculated. Summary relative risk ratios (95% confidence intervals) were calculated. Overall sensitivity was higher for exercise ECG and (quantitatively analysed) MPI than for SE (83.4% and 88.5% versus 74.6% respectively, p<0.0001). SE had a higher specificity (88.7%) than MPI (41.2%) and exercise ECG (60.1%) (p<0.0001). Based on analysis of eight reports, the relative risk of cardiac death or myocardial infarction in patients with an abnormal SE and MPI was elevated more than sevenfold, but it did not differ by imaging modality (p=0.9). Meta-analysis of non-invasive CAD assessment in LBBB patients revealed that exercise ECG and MPI had the highest sensitivity, while SE had the highest specificity. The prognostic accuracy of MPI and SE appeared similar.