Changes in the left ventricular ejection fraction (LVEF) of patients with cardiac sarcoidosis (CS) have been reported; however, the prognostic implications remain unknown. This study aimed to investigate the predictive factors for LVEF improvement in patients with CS and the association between LVEF improvement and prognosis. This was a post hoc analysis of the ILLUMINATE-CS registry, a multicenter retrospective study on CS. Patients with a baseline LVEF ≥ 50% were excluded. LVEF improvement was defined as the difference between the baseline and follow-up echocardiographic results. Patients were stratified based on tertiles of LVEF improvement: 1st (≤ 0%, least improved), 2nd (0-9%), and 3rd (> 9%, most improved) tertile groups. The primary outcome was all-cause mortality after follow-up echocardiography. Overall, 188 patients with CS (age: 62.1 ± 10.7 years; male proportion: 42.6%) were analyzed. In the multivariate linear regression analysis, narrower QRS duration, lower baseline LVEF, and no prescription of beta-blockers at baseline were independently associated with greater LVEF improvement. During the median follow-up period of 407 days, 26 deaths occurred. The Kaplan-Meier curves showed a significant difference between the 3 groups (log-rank test, P = 0.002). In the adjusted Cox proportional hazard analysis, the 1st and 2nd tertile groups showed progressively higher mortality rates than the 3rd tertile group. Independent of other variables, LVEF improvement as a continuous variable was associated with a lower mortality rate. Among patients with CS, LVEF improvement was significantly associated with favorable outcomes. LVEF should be monitored if baseline LVEF is decreased.