Aim To assess the relative value of a novel measure of left ventricular (LV) dyssynchrony derived from magnetic resonance imaging (MRI) and the extent of scar tissue for prediction of response to cardiac resynchronization therapy (CRT). Methods and results Thirty-five heart failure patients scheduled for CRT were included. Left ventricular dyssynchrony was defined as the standard deviation of 16 segment time-to-maximum radial wall thickness (SDt-16) obtained from a cine-set of short-axis slices. Detayed-enhanced MRI was performed for scar analysis. Echocardiography was used to determine response to CRT (reduction >= 15% in LV end-systotic volume 6 months after implantation). At follow-up, 21 patients (60%) were classified as responders. On MRI, SDt-16 was significantly higher in responders compared with non-responders (median 97 vs. 60 ms, P < 0.001), whereas the total extent of scar was larger in non-responders (median 35% vs. 3% in responders, P < 0.001). At the logistic regression analysis, SDt-16 was directly associated (OR = 6.3, 95% CI 3.1-9.9, P < 0.001) and the total extent of scar was inversely associated (OR = 0.52, 95% Cl 0.43-0.87, P < 0.001) with response to CRT. Conclusion Magnetic resonance imaging offers the unique opportunity to assess LV dyssynchrony and scar extent in a single session. Both these parameters are important predictors of echocardiographic response to CRT.