The role of prognostic risk scores in predicting the competing risk of non-sudden death in heart failure patients with reduced ejection fraction (HFrEF) receiving an implantable cardioverter-defibrillator (ICD) is unclear. To this goal, we evaluated the accuracy and usefulness of the Meta-Analysis Global Group in Chronic Heart Failure (MAGGIC) score. The present analysis included 1089 HFrEF ICD recipients enrolled in the OBSERVO-ICD registry (NCT02735811). During a median follow-up of 36 months (1st–3rd IQR 25–48 months), 193 patients (17.7%) experienced at least one appropriate ICD therapy, and 133 patients died (12.2%) without experiencing any ICD therapy. The frequency of patients receiving ICD therapies was stable around 17–19% across increasing tertiles of 3-year MAGGIC probability of death, whereas non-sudden mortality increased (6.4% to 9.8% to 20.8%, p < 0.0001). Accuracy of MAGGIC score was 0.60 (95% CI, 0.56–0.64) for the overall outcome, 0.53 (95% CI, 0.49–0.57) for ICD therapies and 0.65 (95% CI, 0.60–0.70) for non-sudden death. In patients with higher 3-year MAGGIC probability of death, the increase in the competing risk of non-sudden death during follow-up was greater than that of receiving an appropriate ICD therapy. Results were unaffected when analysis was limited to ICD shocks only. The MAGGIC risk score proved accurate and useful in predicting the competing risk of non-sudden death in HFrEF ICD recipients. Estimation of mortality risk should be taken into greater consideration at the time of ICD implantation.