Previously undescribed morphological changes are reported in two cases of recent, non-septic cerebral embolism. These consisted of an arterial rupture involving the whole thickness of the wall. A direct tear by a sharp embolus seems unlikely but, as proposed by Villaret et al. in 1937, the arterial spasm may have occurred at the time of cerebral embolism. The embolic occlusion would prevent narrowing of the vascular lumen, so that focal spasm would rupture the arterial wall even at the level of a healthy and relatively thick media.