Deep vein thrombosis (DVT) remains common in patients with acute ischemic stroke (AIS) receiving aspirin and graded compression stockings (considered standard thromboprophylaxis in the UK), most events occurring in patients with Barthel indices (BI) of <9 ('severe stroke') around the time of admission. In the absence of data indicating improved clinical outcomes with use of low-dose anticoagulant thromboprophylaxis, we evaluated the hypothesis that plasma D-dimers (D-d) might be a valuable initial screening test for proximal DVT (PDVT), facilitating selective use of imaging. One hundred and two unselected AIS-patients receiving aspirin/graded compression stockings thromboprophylaxis were screened for DVT using magnetic resonance direct thrombus imaging, a highly accurate non-invasive technique which directly visualizes thrombus. D-d (VIDAS and IL test D-d assays) were measured on recruitment and at weekly intervals. Median D-d were significantly higher throughout the study in patients with severe stroke who developed PDVT vs. those with severe stroke not developing PDVT, differences being most marked around day 9. Depending on the discriminatory threshold used, a single D-d measurement at this time in patients with severe-AIS allowed identification of a subgroup with ~50% prevalence of PDVT. Sensitivity of these strategies for PDVT was 67-83%, at a cost of imaging 22-30% of the entire cohort of patients. A single measurement of BI around the time of admission and D-d level at day 9 in AIS-patients receiving aspirin/graded compression stockings thromboprophylaxis allows identification of a subgroup containing a substantial proportion of all PDVTs who could be selectively imaged.