For the 183 of 1276 patients in the NINDS Stroke Data Bank with convexity infarction in the middle cerebral artery territory, the size of the infarct did not differ between the two sides but the location of the main site of the infarct differed: on the left side, it was centred in the inferior parietal region, and was mid-frontal on the right. There was a good correlation between infarct size and weakness severity whether estimated by overall motor function on one side, arm, or hand alone. There was a poor correlation, however, for lesion location (lower third, middle third or upper third on either side of the Rolandic fissure) and any of the specific syndromes of focal weakness, no two cases sharing the same lesion for the same syndrome and several cases sharing the same lesion with a different syndrome. The findings indicated a difference in weakness syndromes between the two hemispheres and great individual variation of the acute syndrome caused by a given site of focal infarction along the Rolandic convexity. These variations may explain some of the difficulties showing effects of a given therapeutic agent in studies of acute ischaemic stroke. Large sample sizes will be required for the reliable assessment of any treatment using currently popular clinical stroke scales.