<b><i>Background:</i></b> The newly proposed transient ischemic attack (TIA) definition demands for MRI exclusion of infarction. Due to limited resources other tools than MRI predicting tissue infarction would be valuable. We hypothesized that ABCD² risk score is a valid screening tool for diffusion-weighted imaging (DWI) lesions. <b><i>Methods:</i></b> TIA patients were prospectively enrolled in an observational MRI study to receive acute and follow-up stroke MRI. ABCD² scores were calculated, and sociodemographics and risk factors were recorded. <b><i>Results:</i></b> One hundred and thirty-two TIA patients were enrolled over nine months. Five patients were excluded due to different diagnosis. Forty-five of the 127 remaining patients showed acute ischemic lesions on DWI. Median ABCD² scores for DWI-negative and -positive patients were 4 and 5, respectively. Ordinal, trichotomized and dichotomized ABCD² were significantly associated to DWI. Univariate analysis of single score items and other risk factors demonstrated unilateral weakness, duration of symptoms and smoking as predictive for DWI restrictions. In multivariate analysis unilateral weakness remained significant. <b><i>Conclusions:</i></b> High-risk ABCD² score due to the impact of hemiparesis is associated with the occurrence of DWI lesions but is still not accurate enough for a reliable differentiation of cerebrovascular events with and without MRI lesions.