Objective To test the hypothesis that in recipients of primary prophylactic implantable cardioverter-defibrillators (ICDs), the non-planarity of ECG vector loops predicts (a) deaths despite ICD protection and (b) appropriate ICD shocks.Methods Digital pre-implant ECGs were collected in 1948 ICD recipients: 21.4% females, median age 65 years, 61.5% ischaemic heart disease (IHD). QRS and T wave three-dimensional loops were constructed using singular value decomposition that allowed to measure the vector loop planarity. The non-planarity, that is, the twist of the three-dimensional loops out of a single plane, was related to all-cause mortality (n=294; 15.3% females; 68.7% IHD) and appropriate ICD shocks (n=162; 10.5% females; 87.7% IHD) during 5-year follow-up after device implantation. Using multivariable Cox regression, the predictive power of QRS and T wave non-planarity was compared with that of age, heart rate, left ventricular ejection fraction, QRS duration, spatial QRS-T angle, QTc interval and T-peak to T-end interval.Results QRS non-planarity was significantly (p<0.001) associated with follow-up deaths despite ICD protection with HR of 1.339 (95% CI 1.165 to 1.540) but was only univariably associated with appropriate ICD shocks. Non-planarity of the T wave loop was the only ECG-derived index significantly (p<0.001) associated with appropriate ICD shocks with multivariable Cox regression HR of 1.364 (1.180 to 1.576) but was not associated with follow-up mortality.Conclusions The analysed data suggest that QRS and T wave non-planarity might offer distinction between patients who are at greater risk of death despite ICD protection and those who are likely to use the defibrillator protection.