Background Stroke unit is the most effective treatment method to benefit stroke patients. Our study is to evaluate the early effectiveness of a hospital stroke unit (SU). Methods Three hundred and ninety-two patients who had suffered from acute strokes and who were admitted to our hospital between December 2001 and January 2003 were recruited for this controlled study. All patients were sent at random to either the SU or the general ward (GW) for treatment. The following indices were measured by: Barthel Index (BI), National Institute of Health Stroke Scale (NIHSS), Oxford Handicap Scale (OHS). Results The mean change in BI score between the day of admission and the day of discharge was 20.00 +/- 24.36 for the SU group and 10.63 +/- 23.59 for the GW group. A difference that is statistically significant (P = 0.000). The mean change in NIHSS score was -2.01 +/- 6.61 for the SU group and 0.55 +/- 7.44 for the GW group. A difference that is also statistically significant ( P = 0.000). Finally, the mean change in OHS score was -0.74 +/- 1.04 for the SU group and -0.28 +/- 0.98 for the GW group, also a statistically significant difference ( P = 0.000). Among SU patients, patient satisfaction was higher ( P = 0.000), the rehabilitation success rate was higher (P = 0.000) and there were fewer complications (P = 0.000). Conclusion Compared to GW patients, stroke patients treated in a special SU were able to return to normal daily activities earlier, with better social abilities, and have reduced neurological defects, without increasing the overall economic burden.