Homocysteine increases in the acute phase of ischaemic stroke and from the acute to the convalescent phase, suggesting that hyper-homocysteinaemia may be a consequence rather than a causal factor. Therefore we measured homocysteine plasma levels in stroke patients in order to investigate possible correlations of homocysteine with stroke severity and clinical outcome. Further we looked for eventual differences in stroke subtypes. We prospectively studied plasma homocysteine levels in acute stroke patients admitted to the stroke unit of our department. Seven hundred and seventy-five ischaemic stroke patients, 39 cerebral haemorrhages and 421 healthy control subjects have been enrolled. Stroke severity and clinical outcome were measured with the Scandinavian Stroke Scale, the Rankin Scale and the Barthel Index. Stroke severity by linear stepwise regression analysis was not an independent determinant of plasma homocysteine levels. Homocysteine was not correlated with outcome measured by the Barthel Index. Mean plasma homocysteine of both ischaemic and haemorrhagic stroke was significantly higher than controls (p<0.05). Homocysteine had an adjusted odds ratios (OR) of 4.2 (95% CI 2.77-6.54) for ischaemic stroke and of 3.69 (95% CI 1.90-7.17) for haemorrhagic stroke. Compared with the lowest quartile, the upper quartile was associated with an adjusted OR of ischaemic stroke due to small artery disease of 17.4 (95% CI 6.8-44.3). Homocysteine in the acute phase of stroke was not associated with stroke severity or outcome. Elevated plasma homocysteine in the acute phase of stroke was associated with both ischaemic and haemorrhagic stroke. Higher levels are associated with higher risk of small artery disease subtype of stroke.