The purposes of the present study were to determine the difference of functional outcome, and to identify factors associated with functional difference in patients with stroke. The sample consisted of 56 stroke patients who had received physical therapy at the physical therapy unit of the Dongeui Medical Center in Busan city between January 2000 and June 2002. Stroke patients were evaluated by physical therapists 3 times; The first day in physical therapy (PT) (T1), one month after the first day in PT (T2), and two months after the first day in PT (T3). Functional status was assessed with the Functional Independence Measure (FIM) instrument, a validated instrument for documenting the severity of disability and assessing the outcome of rehabilitation treatment. Functional gain was calculated over T2-T1, T3-T1, and T3-T2. SAS statistical software was used for the analysis. The Student's t-test, paired t-test, analysis of variance (ANOVA/Tukey and Scheffe), and analysis of covariance (ANCOVA) were used to examine the functional difference in variables. Repeated measures ANOVA was also used to analyze the functional difference by time (T1, T2, and T3). Multiple regression analysis was performed to determine the effects of independent variables on the difference of functional outcome as defined by the FIM score. A total of 56 stroke patients were evaluated, their average age±standard deviation was years (range: 40~81 yr). The functional status of patients who received physical therapy for about 2~3 months was significantly improved (mean FIM scores, 20.5±1.8, 28.9±1.9, and 8.41±1.1 points for each time period, respectively) (p<.0001). Diabetes was significantly associated with the FIM score for T2-T1 (p<.05). The type of diagnosis was significantly associated with the FIM score for T3-T1 (p<.05). Gender, smoking, and the FIM score on admission were significantly associated with the FIM score for T3-T2 (p<.05). In conclusion, gender, smoking, diabetes, the type of diagnosis, and the FIM score on admission were significantly associated with improved FIM scores. We recommend that further research should explore the functional outcome by using larger sample sizes, longer follow-up periods, and more sensitive assessment instruments.