Aim <p>A strong need exists to better select patients with reduced left ventricular ejection fraction for primary prevention of sudden cardiac death by ICD implantation. This paper reports on the expected clinical and economic benefits of stricter indication scenarios based on minimum probabilities for patients of experiencing appropriate ICD-therapy and/or maximum risks of dying during the first 2&#xa0;years following ICD implantation.</p> Methods <p>Data on clinical events and hospital care expenditures were gathered for patients in the Dutch DO-IT registry. Registry-based prediction models were used to derive individual prior probabilities. Realistic assumptions were made concerning short-term disease courses and related hospital care in absence of ICD implantation. The potential impact of stricter indication scenarios was assessed with changes in numbers needed to treat (NNT) in subpopulations with (NNT-yes) or without (NNT-no) indication for ICD implantation and with changes in the yearly incidence-based national hospital care budget for ICD-based primary prevention.</p> Results <p>The NNT-yes under the existing guidelines equalled 42. Not indicating ICD implantation if prior probabilities of receiving appropriate therapy within 2-year post-implant are &lt; 5% seems a promising cutoff with an NNT-yes of 33, an NNT-no of 246 and a national annual reduction in hospital expenditures for ICD-based primary prevention of €11 million (16.7%).</p> Conclusions <p>Stricter indication criteria for primary prevention ICD implantation enable the selection of patient subpopulations with high numbers needed to treat, in which unnecessary harm can be forgone and substantial savings can be accomplished. The scenario analysis facilitates rationing of indication policies for ICD implantations.</p> Graphical abstract <p></p>