Background and Purpose – European directives and legislation in some countries forbid inclusion of subjects incapable of consent in research if recruitment of patients capable of consent will yield similar results. We compared brain lesion volumes in stroke patients deemed to have capacity to consent with those defined as incapacitated. Methods – Data were obtained from 3 trials recruiting patients primarily with cortical stroke syndromes. Patients were recruited within 24 hours of onset and used MRI based selection or outcome criteria. Method of recruitment was recorded with stroke severity, age, and brain lesion volumes on Diffusion Weighted Imaging. Results – Of the 56 subjects included, 38 (68%) were recruited by assent and 18 (32%) by consent. The assent group had a median lesion volume of 18.35 cubic centimetres ( cc) ( interquartile range [IQR] 8.27-110.31 cc), compared to 2.79 cc ( IQR 1.31-12.33 cc) when patients consented (P=0.0004). Lesions were smaller than 5 cc in 7/38 (18%) in the assent group and 11/18 (61%) in the consent group (P=0.0024). There was good correlation between neurological deficit by NIH stroke scale score and lesion volume (r=0.584, P < 0.0001). Logistic regression demonstrated NIHSS or lesion volume predicted capacity to consent. Conclusions - Patients with acute stroke who retain capacity to consent have significantly smaller infarct volumes than those incapable of consent, and these are frequently below the limits where measurement error significantly compromises valid use of volumetric end points. Only a small proportion of patients with capacity to consent would be eligible for, and contribute usefully to, most acute stroke trial protocols.