The clinical significance of serum S-100 protein, a protein released by damaged brain tissue, was assessed in patients with acute ischaemic or haemorrhagic stroke and matched controls. Serum S-100 protein concentration was significantly elevated in patients with ischaemic stroke [median (SQR): 0.27 (0.09) microgram/L, n = 68] and haemorrhagic stroke [0.43 (0.23) microgram/L, n = 13] compared to controls [0.11 (0.03) microgram/L, n = 51, P < 0.0001]. Although patients with haemorrhagic stroke had higher serum S-100 concentrations compared to patients with ischaemic stroke, this was not quite statistically significant. Serum S-100 concentrations were related to infarct size, large (total anterior circulation) infarcts concentrations having the highest [0.40 (0.22) microgram/L], and small vessel ('lacunar') infarcts concentrations having the lowest [0.20 (0.06) microgram/L, P < 0.0005] concentrations. S-100 protein concentration was also significantly related to clinical outcome at three months measured using three disability and handicap scales (n = 81): modified Barthel index (rs = -0.285, P = 0.01), modified Rankin score (rs = 0.313, P = 0.004) and Lindley score (rs = 0.262, P = 0.018) with high values associated with poor clinical outcome. Similarly high values of serum S-100 protein were observed in patients who died or were discharged to an institution [median (SQR): 0.63 (0.29) microgram/L and 0.37 (0.13) microgram/L, respectively] compared to those who were discharged home [0.26 (0.11) microgram/L, P = 0.13]. The present study suggests measurement of serum S-100 protein could be a useful prognostic marker of clinical outcome in acute stroke. Whether S-100 concentrations can be altered by therapeutic intervention in acute stroke remains to be elucidated.