Purpose: To assess the association between Veterans Affairs (VA) stroke patients' poststroke rehabilitation utilization and their residential settings by using 2 common rural-urban taxonomies. Methods: This retrospective study included all VA stroke inpatients in 2001 and 2002. Rehabilitation utilization referred to rehabilitation therapy received 12-months poststroke hospitalization. Patients' urban, rural, or isolated/highly rural status was determined using the Rural-Urban Commuting Areas (RUCA) and VA Rural Urban (VARU) definitions based on patient residential ZIP code. Logistic regression models were fit for the rehabilitation outcome, adjusting for potential risk factors. Findings: Among the 8,296 stroke patients, 69.6%/61.1% were categorized as urban, 21.3%/37.5% as rural, and 9.1%/1.4% as isolated/highly rural by the RUCA/VARU definitions, respectively. Compared with their urban counterparts, the rural and/or isolated/highly rural patients were significantly more likely to be older, white, married, living further from the VA hospitals, not hospitalized for stroke directly from home, and not intubated. Compared with the rural patients, odds of receiving rehabilitation therapy were 1.2 times (RUCA) and 1.1 times (VARU) by the urban patients, and 0.53 times (VARU only) by the highly rural patients, after risk adjustment. The above comparisons were significant at P less than 0.05. Conclusions: With both taxonomies, the rural patients were less likely to receive postacute stroke rehabilitant therapy than their urban counterparts. With the VARU, the highly rural patients were less likely to receive rehabilitation care than their rural counterparts. Different taxonomy may lead to different rural-urban classification yields and different yields may lead to different outcomes and conclusions.