BACKGROUND: A small proportion of patients undergoing primary prophylactic implantation of implantable cardioverter defibrillators (ICDs) experiences malignant arrhythmias. We postulated that periodic repolarisation dynamics, a novel marker of sympathetic-activity-associated repolarisation instability, could be used to identify electrically vulnerable patients who would benefit from prophylactic implantation of ICDs by way of a reduction in mortality. METHODS: We did a prespecified substudy of EUropean Comparative Effectiveness Research to Assess the Use of Primary ProphylacTic Implantable Cardioverter Defibrillators (EU-CERT-ICD), a prospective, investigator-initiated, non-randomised, controlled cohort study done at 44 centres in 15 EU countries. Patients aged 18 years or older with ischaemic or non-ischaemic cardiomyopathy and reduced left ventricular ejection fraction (≤35%) were eligible for inclusion if they met guideline-based criteria for primary prophylactic implantation of ICDs. Periodic repolarisation dynamics from 24-h Holter recordings were assessed blindly in patients the day before ICD implantation or on the day of study enrolment in patients who were conservatively managed. The primary endpoint was all-cause mortality. Propensity scoring and multivariable models were used to assess the interaction between periodic repolarisation dynamics and the treatment effect of ICDs on mortality. FINDINGS: Between May 12, 2014, and Sept 7, 2018, 1371 patients were enrolled in our study. 968 of these patients underwent ICD implantation, and 403 were treated conservatively. During follow-up (median 2·7 years [IQR 2·0-3·3] in the ICD group and 1·2 years [0·8-2·7] in the control group), 138 (14%) patients died in the ICD group and 64 (16%) patients died in the control group. We noted a 43% reduction in mortality in the ICD group compared with the control group (adjusted hazard ratio [HR] 0·57 [95% CI 0·41-0·79]; p=0·0008). Periodic repolarisation dynamics significantly predicted the treatment effect of ICDs on mortality (adjusted p=0·0307). The mortality benefits associated with ICD implantation were greater in patients with periodic repolarisation dynamics of 7·5 deg or higher (n=199; adjusted HR 0·25 [95% CI 0·13-0·47] for the ICD group vs the control group; p<0·0001) than in those with periodic repolarisation dynamics less than 7·5 deg (n=1166; adjusted HR 0·69 [95% CI 0·47-1·00]; p=0·0492; pinteraction=0·0056). The number needed to treat was 18·3 (95% CI 10·6-4895·3) in patients with periodic repolarisation dynamics less than 7·5 deg and 3·1 (2·6-4·8) in those with periodic repolarisation dynamics of 7·5 deg or higher. INTERPRETATION: Periodic repolarisation dynamics predict mortality reductions associated with prophylactic implantation of ICDs in contemporarily treated patients with ischaemic or non-ischaemic cardiomyopathy. Periodic repolarisation dynamics could help to guide treatment decisions about prophylactic ICD implantation. FUNDING: The European Community's 7th Framework Programme.