The significance of the stroke unit (SU) in Japan, where the medical system differs from those in Europe and the USA, remains unclear. We conducted a prospective multicenter study to clarify this issue. The study subjects consisted of 7614 consecutive patients with completed stroke, excluding subarachnoid hemorrhage, who were admitted to 117 hospitals in Japan within 72 hours of stroke onset from December 2004 to December 2005. We divided the hospitals into those with and those without an SU, and investigated whether or not the treatment in the SU could improve the patients' clinical outcome. In our preliminary analysis using data for the initial 4268 patients, logistic regression analysis demonstrated that the only significant and independent predictor of case fatality at 28 days and 3 months after onset was the NIHSS score on admission. Treatment in the SU, however, was significantly associated with a favorable outcome (mRS, 0-2) at 3 months after stroke onset as well as age, male gender, the pre-stroke mRS score, and the NIHSS score on admission. Assessments of swallowing functions and the application of an early rehabilitation program within the initial week were more frequently given in acute stroke patients at hospitals with an SU than at those without an SU. The present results indicate that the treatment in an SU may also improve the clinical outcome of patients at 3 months after stroke onset in Japan.