AIMS: We sought to determine risk models for predicting early and late stroke in a large cohort of high-risk post-myocardial infarction (MI) patients. METHODS AND RESULTS: We prospectively analysed data from 14 703 patients in the VALIANT trial with acute MI complicated by heart failure, left ventricular (LV) systolic dysfunction, or both. Patients were randomized 0.5-10 days after acute MI to valsartan, captopril, or their combination. We evaluated risk factors for early (<45 days) and late (>45 days) stroke by using multivariable Cox proportional hazards regression analyses with stepwise variable selection techniques applied to 92 pre-specified potential predictor variables. After randomization, 463 (3.2%) patients had fatal (n = 124) or non-fatal (n = 339) strokes, with 134 strokes occurring in the first 45 days. The strokes were classified as ischaemic (348), haemorrhagic (40), or of indeterminate cause (75). Estimated glomerular filtration rate and heart rate when in sinus rhythm were the most powerful predictors of early stroke (<45 days after MI), whereas diastolic blood pressure (DBP) >90 mmHg, prior stroke, and atrial fibrillation (AF) were the most powerful predictors of stroke overall. Ejection fraction and sex were not predictive of stroke in this cohort. CONCLUSION: Among high-risk patients presenting with MI but without initial neurological symptoms, the risk of stroke 6 weeks thereafter is 0.94% (95% CI 0.78-1.09). Of the most powerful baseline predictors of stroke, DBP and AF are amenable to therapeutic interventions and thus merit special attention in these patients.