Abstract Background There is still a big unmet need for better risk stratification tools to identify patients at high risk for sudden cardiac death in the era of primary prevention ICD therapy. Scar burden on late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) imaging may be a risk marker for the prediction of ventricular arrhythmia in post-myocardial infarction (MI) patients. Purpose The aim of the PARCADIA trial was to assess whether scar characteristics, determined via LGE-CMR, or other baseline risk markers are associated with appropriate ICD therapy. Methods In this prospective, multi-center, clinical trial, we enrolled post-MI patients with an indication for de novo primary prevention ICD. Main exclusion criteria were indication for cardiac resynchronization therapy and NYHA class IV. Prior to the ICD implantation, LGE-CMR was performed to analyze the amount of scar tissue (LGE core mass + grey zone mass) and LV function / volumes. A 24h Holter recording was performed to collect information on potential triggers like premature ventricular beats (PVB) and sympathetic-vagal balance based on heart rate variability (HRV). Follow-ups were performed in-office every 6 months, including remote monitoring. Appropriate ICD therapy was used as primary endpoint and adjudicated by a clinical event committee. Results 199 patients were enrolled at 5 investigational sites in Europe. Patients mean age was 64.3 years, with a mean LVEF of 27.7 %, timing from the index MI was 9.1 years, and 42.7% received a primary PCI on index MI. After a median follow-up duration of 47.7 months, 23.1% of patients received appropriate ICD therapy. In 135 patients (67.8%), quality of LGE-CMR was sufficient for detailed analysis. The mean LGE core mass (20.5 vs. 19,7 g) and grey zone mass (10.5 vs. 11.4 g) was not different in patients with or without appropriate ICD therapy, also not when corrected for LV mass. However, LVEF (28.5 vs. 32.9 %) was lower, while diastolic myocardial mass (141 vs. 125 g), and end-diastolic volume (265.0 vs. 226.7 ml) were significantly higher in patients with appropriate ICD therapy. On 24h Holter the mean PVB burden (3695 vs. 1643 beats), and non-sustained ventricular tachycardia (12 vs. 0.9), were higher in patients with appropriate ICD therapy, while the HRV (96.1 vs. 113.1 SDNN) was lower. In univariate analysis the variables LVEF, timing to index MI, primary PCI on index MI, PVB burden and HRV appeared to be predictors for appropriate ICD intervention, while in multivariate analysis LVEF and PVB burden remained predictive. Conclusion In post-MI patients with appropriate ICD therapy, scar burden was not different from patients without appropriate ICD therapy. Only LVEF and PVB burden were independent predictors of increased arrhythmogenicity. Improved CMR analysis on scar characteristics (conduction channels) to identify potential arrhythmogenic regions might be the topic for future research.