Objective : Despite improvement of therapeutic regimen, incidence of stroke increases and it remains a leading cause of death. Our study aims at offering variable data on recurrent strokes. Methods : There were 59 patients who admitted from Jan. 2002 to Dec. 2004 due to recurrent strokes. A retrospective longitudinal cohort study was done. Results : Four-hundred-seventy five patients, diagnosed with acute stroke, experienced 491 strokes in 3 years, and there were 75 recurrent strokes [15.3%] in 59 patients. These 59 patients were included in the study. First hemorrhagic cases [H] were 19 [32%], and the first infarction cases [I] were 40 [68%]. Subsequent strokes after first stroke were as follows : $H{ o}H$ 14 [23.7%] cases, $H{ o}I$ 5 [8.5%], $I{ o}H$ 8 [13.6%], $I{ o}I$ 32 (54.2%]. A Cox regression analyses showed that the first type of stroke was a significant factor to the second stroke as follows : if one has had a hemorrhagic stroke, the possibility of second hemorrhagic attack ($H{ o}H$ attack) increase 3.2 times than ischemic type and in ischemic stroke [$I{ o}I$ attack] 3.6 times increased incidence of second ischemic attack. Conclusion : The recurrence rate of stroke was 12.4% [59 of 475 patients]. If the first stroke is hemorrhage or infarction, the next stroke would have high potentiality of hemorrhage, or infarction. The possibility of same type in second stroke Increase over 3 times. In $H{ o}H$ group, the time interval between first and second stroke was shorter and the age of onset was earlier than in $I{ o}I$ group. Moreover, the infarction was more frequent than hemorrhage in multiple strokes. There was a correlation in lacunar type infarction between first and second attack.