To date, there is no data regarding the association of patient in-hospital referral source and stroke rehabilitation outcomes. The objective of the present study was to investigate the possible relation between in-hospital referring source, whether directly from an emergency ward (EW) or indirectly through a general medical ward (GMW), and the functional outcomes achieved during rehabilitation of such stroke patients. This retrospective observational study included 315 consecutive patients, admitted for rehabilitation following the onset of acute stroke. We compared those referred directly to us from the EW, with others referred from GMWs. Functional status was assessed by Functional Independence Measure method (FIM). Functional outcome was determined by total FIM gain (efficacy) and daily FIM gain (efficiency), both absolute and relative (to potential). The two study groups were similar in terms of age, gender, and diagnosis. FIM admission scores were higher at admission in patients admitted directly from the EW, compared with those referred from GMWs (72.5 +/- 27.5 and 62.7 +/- 25.6, respectively) but similar at discharge (77.4 +/- 28.8 and 80.7 +/- 32.5, respectively). Length of stay (LOS) in the GMW group was longer as compared to the EW group. Efficacy was significantly associated with being married, younger age, hemiplegia, and admission scores between 40-60. Both absolute and relative efficacy and efficiency rates of rehabilitation were significantly lower among patients referred from the EW. We conclude that in-hospital referral source is associated with different rehabilitation outcomes in stroke patients. Direct admission of stroke patients from the EW is associated with lower rehabilitation efficacy and efficiency rates, compared with those admitted from GMWs. The findings support the implementation of different selection methods, underscoring the need of both clinicians and administrators to consider the in-hospital referral source as a potential factor associated with stroke rehabilitation outcome.