Abstract Background The number of centers with left ventricular assist device (LVAD) research programs focused on cardiac recovery is very small. Therefore, this phenomenon has been reported in real-world multi-center registries as a rare event. Objectives This study evaluated the incidence of cardiac recovery with an a priori LVAD implantation strategy of bridge-to-recovery (BTR) and constructed a recovery predictive model. Methods The study included LVAD recipients registered in the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS). Cardiac recovery was evaluated in BTR and non-BTR patients. A weighted score was derived and externally validated in patients of the Utah Cardiac Recovery (UCAR) program. Results Of 15,138 INTERMACS patients, cardiac recovery occurred in 192 (1.3%). The incidence of recovery was 11.2% (n = 14) in BTR compared with 1.2% (n = 178) in non-BTR patients (p < 0.0001). Independent predictors of recovery included: age <50 years, non-ischemic cardiomyopathy, time from cardiac diagnosis <2 years, absence of ICD, creatinine ≤1.2 mg/dl, and LVEDD <6.5 cm (c-index: 0.85; p < 0.0001). A weighted score termed I-CARS, effectively stratified patients based on their probability of recovery. I-CARS was validated in the UCAR cohort (n = 190) with good performance (AUC: 0.94; 95% CI: 0.91 to 0.98). One-year survival after LVAD explantation, available in INTERMACS for 21 (11%) patients, was 86%. Conclusions The incidence of cardiac recovery is higher in patients implanted with an a priori BTR strategy. We developed a simple tool to help identify patients in whom recovery is feasible. In BTR patients with favorable characteristics, I-CARS suggests a 24% probability of successful LVAD explantation. Large-scale studies to better address post-explantation outcomes are warranted.