BACKGROUND: In patients with acute myocardial infarction (MI), BMI < 18.5 kg/m2 and a decrease in BMI during follow-up have been associated with poor prognosis. For BMI ≥ 25 kg/m2, an "obesity paradox" has been suggested. Recently, high visit-to-visit BMI variability has also been associated with poor prognosis in patients with coronary artery disease. AIMS: To simultaneously evaluate several BMI measurements and study their association with cardiovascular (CV) outcomes in a large cohort of patients with acute myocardial infarction (MI) and left ventricular (LV) systolic dysfunction, heart failure (HF) or both. METHODS: The high-risk MI dataset is pooled from four trials: CAPRICORN, EPHESUS, OPTIMAAL and VALIANT. Mean BMI, change from baseline, and variability were assessed during follow-up. The primary outcome was CV death. Cox-proportional hazard models were performed to study the association between the various BMI parameters and outcomes (median follow-up = 1.8 years). RESULTS: A total of 12,719 patients were included (72% male, mean age 65 ± 11 years). Mean, change and visit-to-visit variability in BMI had a non-linear association with CV death (P < 0.001). Mean BMI < 26 kg/m2 (vs. ≥ 26-35 kg/m2) and BMI decrease during follow-up were independently associated with CV death (adjusted HR 1.32, 95% CI 1.16-1.51, P < 0.001 and adjusted HR 1.57, 95% CI 1.40-1.76, P < 0.001, respectively). Low and high BMI variability (< 2% and > 4%) were associated with increased event-rates, but lost statistical significance in sensitivity analysis including patients with ≥ 5 measurements or excluding patients with HF hospitalization, suggesting that BMI variability may be particularly associated with HF hospitalizations. CONCLUSION: Mean BMI < 26 kg/m2 and a BMI decrease during follow-up were independently associated with CV death in patients with MI and LV systolic dysfunction, HF or both. These associations likely reflect poorer patient status and causality cannot be inferred.