<p><b>BACKGROUND</b>Although cardiac resynchronization therapy (CRT) is already an established treatment, the characteristics of patients who have an excellent response to CRT and those who get no benefit remain to be determined. The purpose of this study was to search for potential predictors of both non-response and super-response to CRT.</p><p><b>METHODS</b>Seventy-six consecutive patients who received CRT treatment were divided into group A (non-responders), group C (super-responders) and group B (responders exclusive of super-responders). Student's t test, Mann-Whitney test, Logistic regression and receiver operating characteristic curve were employed to identify potential predictors among the patients' demographic characteristics, clinical features, several electrocardiographic parameters before and after CRT implantation, and their pre-implant echocardiographic parameters.</p><p><b>RESULTS</b>Group A had the lowest 3-month left ventricular ejection fraction (LVEF). Group C had the smallest pre-implant left ventricular end-diastolic dimension (LVEDD), the shortest post-implant QRS duration, the smallest 3-month LVEDD and the highest 3-month LVEF. In addition, there was a trend of gradual change in percent of left bundle branch block, severity of pre-implant mitral regurgitation, pre-implant QRS dispersion, post-implant QRS duration as well as post-implant QRS dispersion from group A to group B and from group B to group C. Multivariable Logistic analysis revealed that only pre-implant LVEDD could predict CRT super-response. A pre-implant LVEDD of 68.5 mm was the cut-off value that identified super-responders with 87.5% sensitivity and 79.7% specificity. A pre-implant LVEDD of 62.5 mm identified super-responders with 50.0% sensitivity and 89.8% specificity.</p><p><b>CONCLUSIONS</b>Predictors of a CRT non-response remain unclear at present. But it is credible that patients with a smaller left ventricle would have a better chance to become super-responders to CRT.</p>