Impaired glucose tolerance is often present in patients with a transient ischemic attack (TIA) or ischemic stroke and doubles the risk of recurrent stroke. This impaired glucose tolerance can be transient, reflecting an acute stress response, orpersistent, representing undiagnosed impaired glucose metabolism possibly requiring treatment. We aimed to assess the occurrence of persistent impaired glucose tolerance after a stroke or TIA and to develop a prediction model to identify patients at risk of persistent impaired glucose tolerance. Patients admitted to the stroke unit or TIA clinic of the Erasmus Medical Center with ischemic stroke or TIA and impaired glucose tolerance (2-hour postload glucose level of 7.8-11.0mmol/L) were consecutively enrolled between July 2009 and June 2012. The oral glucose tolerance test was repeated after 3months and patients were classified as having transient impaired glucose tolerance or persistent impaired glucose tolerance. We developed a prediction model by means of a multivariable logistic regression model. We calculated the area under the receiver operating characteristic curve (AUC) to quantify the performance of the model and the internal validity by bootstrapping. Of the 101 patients included, 53 (52%) had persistent impaired glucose tolerance or progression to diabetes. These patients were older and more often had hypertension and used statins. A prediction model including age, current smoking, statin use, triglyceride, hypertension, previous ischemic cardiovascular disease, body mass index, and fasting plasma glucose accurately predicted persistent impaired glucose tolerance (bootstrapped AUC, .777), with statin use, triglyceride, and fasting plasma glucose as the most important predictors. Half of the patients with impaired glucose tolerance after a TIA or ischemic stroke have persistent impaired glucose tolerance. We provide a prediction model to identify patients at risk of persistent impaired glucose tolerance, with statin use, triglyceride, and fasting plasma glucose as the most important predictors, which after external validation might be used to optimize secondary prevention.