Abstract Aims Patient selection for cardiac resynchronization therapy (CRT) may be enhanced by evaluation of systolic myocardial stretching. We evaluate whether systolic septal rebound stretch (SRSsept) derived from speckle tracking echocardiography is a predictor of reverse remodelling after CRT and whether it holds additive predictive value over the simpler visual dyssynchrony assessment by apical rocking (ApRock). Methods and results The association between SRSsept and change in left ventricular end-systolic volume (ΔLVESV) at 6 months of follow-up was assessed in 200 patients. Subsequently, the additive predictive value of SRSsept over the assessment of ApRock was evaluated in patients with and without left bundle branch block (LBBB) according to strict criteria. SRSsept was independently associated with ΔLVESV (β 0.221, P = 0.002) after correction for sex, age, ischaemic cardiomyopathy, QRS morphology and duration, and ApRock. A high SRSsept (≥optimal cut-off value 2.4) also coincided with more volumetric responders (ΔLVESV ≥ −15%) than low SRSsept in the entire cohort (70.0% and 56.4%), in patients with strict LBBB (83.3% vs. 56.7%, P = 0.024), and non-LBBB (70.7% vs. 46.3%, P = 0.004). Moreover, in non-LBBB patients, SRSsept held additional predictive information over the assessment of ApRock alone since patients that showed ApRock and high SRSsept were more often volumetric responder than those with ApRock but low SRSsept (82.8% vs. 47.4%, P = 0.001). Conclusion SRSsept is strongly associated with CRT-induced reduction in left ventricular end-systolic volume and holds additive prognostic information over QRS morphology and ApRock. Our data suggest that CRT patient selection may be improved by assessment of SRSsept, especially in the important subgroup without strict LBBB. Clinical trial registration The MARC study was registered at clinicaltrials.gov: NCT01519908.