Introduction: Guidelines favor patient selection by left bundle branch block (LBBB) with QRS width ≥150 ms for cardiac resynchronization therapy (CRT). Predicting CRT response in patients with QRS width 120 to 149 ms or non-LBBB remains difficult. Left ventricular (LV) global longitudinal strain (GLS) and systolic stretch index (SSI) have shown to characterize the ventricular substrate responsive to CRT. However, the potential application of longitudinal left atrial (LA) strain remains unclear. Hypothesis: Baseline LA strain has prognostic value in CRT patients with intermediate ECG criteria. Methods: We studied 195 patients who underwent CRT based on routine indications: ejection fraction ≤35% and QRS width ≥120 ms. GLS was assessed using the 3 standard apical views. Radial SSI was derived from the mid LV short axis view. Peak longitudinal LA strain was derived from the 2 and 4-chamber apical view. The predefined combined clinical endpoint was death, heart transplant or left ventricular assist device over 4 years after CRT. Results: LA strain was feasible in 162 (83%) of the CRT candidates: QRS duration 156 ± 26 ms, 39.5% had LBBB with QRS ≥ 150ms, 60.5% had intermediate ECG criteria. High peak longitudinal strain (>median, 10.1%) was associated with favorable event-free survival (p<0.001). Patients with intermediate ECG criteria for CRT and both high peak longitudinal LA strain and high GLS (>median, 8.4%) had similar outcome to those with Class I indications for CRT. Multivariable analysis revealed that LA strain had independent prognostic value (hazard ratio 0.9 per LA strain %, p < 0.001) even after adjusted for other clinical, electrophysiological and echocardiographic covariates including QRS morphology and duration, GLS and SSI. Conclusions: Peak LA strain had important prognostic value in candidates for CRT. Prognostic value of LA strain was additive to LV strain characteristics and most significant in CRT patients with intermediate ECG criteria for CRT.