Aims In a time of controversy regarding the use of echocardiography for assessment of left ventricular (LV) dyssynchrony and prediction of response to cardiac resynchronization therapy (CRT), this meta-analysis aimed to evaluate the feasibility and reliability of LV dyssynchrony assessment by three-dimensional echocardiography (3DE), determine clinically useful reference values in healthy subjects and heart failure patients, and examine the accuracy of 3DE to predict response to CRT. Methods and results A total of 73 studies that evaluated the assessment of LV dyssynchrony by 3DE were eligible. The systolic dyssynchrony index (SDI) for 16 segments, being the predominant 3DE dyssynchrony parameter, was used for data pooling. Results demonstrated that LV dyssynchrony assessment by 3DE is feasible in 94% of studied subjects [95% confidence interval (CI): 92-95%)]. Pooled estimates of intraclass correlation coefficients (ICC) and limits of agreement (LoA) demonstrated that SDI has good interobserver (ICC: 0.92, LoA: 4.07%) and intraobserver reliability (ICC: 0.95, LoA: 2.10%). Reference values of SDI in healthy subjects, heart failure patients in general, and patients eligible for CRT were 2.7±0.9%, 9.8±3.9%, and 10.7±3.6%, respectively. Meta-regression analysis demonstrated that reference values of SDI in healthy subjects significantly differed between different software [1.80% (95% CI: 0.55-3.05%), P < 0.001]. In patients eligible for CRT, SDI had good accuracy to predict treatment response with a weighted mean cut-off value of 9.8% and pooled estimates for sensitivity and specificity of 93% (95% CI: 89-97%) and 75% (95% CI: 58-93%), respectively. Conclusion 3DE is a feasible and reliable tool for assessment of LV dyssynchrony and may have additional value to current selection criteria for accurate prediction of response to CRT. © The Author 2012.