Rupture of a vulnerable carotid plaque is an important cause of ischemic stroke. Prediction models can support medical decision-making by estimating individual probabilities of future events, while magnetic resonance imaging (MRI) can provide detailed information on plaque vulnerability. In this review, prediction models for medium to long-term (>90 days) prediction of recurrent ischemic stroke among patients on best medical treatment for carotid stenosis are evaluated, and the emerging role of MRI of the carotid plaque for personalized ischemic stroke prediction is discussed. A systematic search identified two models; the European Carotid Surgery Trial (ECST) medical model, and the Symptomatic Carotid Atheroma Inflammation Lumen stenosis (SCAIL) score. We critically appraised these models by means of criteria derived from the CHARMS (CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modeling Studies) and PROBAST (Prediction model Risk Of Bias ASsessment Tool). We found both models to be at high risk of bias. The ECST model, the most widely used model, was derived from data of large but relatively old trials (1980s and 1990s), not reflecting lower risks of ischemic stroke resulting from improvements in drug treatment (e.g., statins and anti-platelet therapy). The SCAIL model, based on the degree of stenosis and positron emission tomography/computed tomography (PET/CT)-based plaque inflammation, was derived and externally validated in limited samples. Clinical implementation of the SCAIL model can be challenging due to high costs and low accessibility of PET/CT. MRI is a more readily available, lower-cost modality that has been extensively validated to visualize all the hallmarks of plaque vulnerability. The MRI methods to identify the different plaque features are described. Intraplaque hemorrhage (IPH), a lipid-rich necrotic core (LRNC), and a thin or ruptured fibrous cap (TRFC) on MRI have shown to strongly predict stroke in meta-analyses. To improve personalized risk prediction, carotid plaque features should be included in prediction models. Prediction of stroke in patients with carotid stenosis needs modernization, and carotid MRI has potential in providing strong predictors for that goal.