Background: Previous studies in the setting of patients with acute myocardial infarction (AMI) have demonstrated that hypovitaminosis D is associated with increased mortality risk during a follow-up whose median did not exceed two years. Objective: To evaluate the impact of vitamin D levels on long-term mortality in patients with AMI. Results: In our study 477 patients with AMI were included. During a median follow-up period of 57 (IQR 53–64) months, 93 patients (20%) died. A non-linear U-shaped relationship between 25(OH)D levels and long-term mortality was observed; patients with vitamin D b 10 ng/mL and N30 ng/mL had higher mortality rate than those with intermediate values. After adjustment for differences in baseline features and treatment, it was con- firmed that extreme values of vitamin D (b10 or N30 ng/mL) are independent predictors of mortality with HR of 3.02 (95% CI 1.78–5.11). Other independent predictors of outcome were age, NYHA class at discharge, treatment with ACE inhibitors and statins. The estimated time-dependent ROC curve of the multivariable model including vitamin D showed an AUC significantly higher than the model without vitamin D: AUC 0.82 (95% CI 0.76–0.87) vs. 0.77 (95% CI 0.71–0.83), p = 0.005. Addition of vitamin D to the model that included all significant factors for mortality improved the prognostic accuracy as showed by the metrics of reclassification (NRI 0.34 (95% CI 0.14–0.48), p = 0.003 and IDI 0.06 (95% CI 0.01–0.12, p = 0.005 p = 0.03). Conclusions: We report a U-shaped relationship between vitamin D levels and long-term outcome of patients surviving AMI.