OBJECTIVE To explore racial/ethnic differences in FIM data from admission to discharge in underinsured patients undergoing inpatient stroke rehabilitation. METHODS This is a retrospective analysis of the Uniform Data System for Medical Rehabilitation (UDSMR) database of an inpatient rehabilitation unit of a county hospital in a large urban city. Data included 171 adult patients admitted to the stroke rehabilitation unit between January 2000 and October 2003. Main outcome measures included admission and discharge total FIM score, FIM gain, FIM efficiency, and length of stay (LOS). Data were analyzed using chi analyses, t tests, univariate analysis of variance, binary logistic regression, and hierarchical multiple regression. RESULTS Data from 68 Hispanic, 83 black, and 20 white patients were included in the study. Univariate tests revealed that race/ethnicity groups differed significantly on admission FIM score (F=5.38, P<0.005), FIM gain (F=4.35, P<0.014), and FIM efficiency (F=3.42, P<0.035). Post hoc pairwise comparisons revealed that Hispanics had lower admission FIM scores than blacks (58.9 vs. 68.9). However, Hispanics had higher FIM gain scores than blacks (26.8 vs. 21.5). Race/ethnicity was not significantly related to age, gender, side of stroke, type of stroke, time from onset of stroke to rehabilitation admission, discharge FIM score, or LOS. Multiple regression analyses revealed that after controlling for all other available factors, race/ethnicity accounted for a significant amount of additional variance in admission FIM score (5.8%) and FIM efficiency (4.6%), but not in discharge FIM score, FIM gain, or LOS. Race/ethnicity was not predictive of discharge disposition. CONCLUSIONS Differences in functional independence at admission to poststroke rehabilitation and the average daily improvement in function are related, in part, to patients' race/ethnicity. Differences in change in functional independence from admission to discharge (FIM gain) are not related to race/ethnicity once other factors, particularly admission FIM score, are taken into account. Future studies should identify reasons why Hispanics have lower admission FIM scores because demographic and stroke-related variables were not related to ethnicity yet have outcomes similar to blacks and whites at discharge.