The objective of this study was to identify the variables associated with discharge to inpatient rehabilitation following acute severe stroke and to determine whether hospital unit contributed to access. Five acute hospitals in Victoria, Australia participated in this study. Patients were eligible for inclusion if they had suffered an acute severe stroke (Mobility Scale for Acute Stroke [less than or equal to] 15). Physiotherapists assessed patients on day 3 poststroke, collecting demographic information and information relating to their prestroke status, social status and current status. Stepwise logistic-regression modelling was used to examine the association between age, type of stroke, prestroke living situation, comorbidities, availability of carer on discharge, current mobility, bladder continence, bowel continence, cognition and communication and the dependent variable, discharge destination (rehabilitation/other). The resulting model was analysed using hierarchical logistic regression with hospital unit as the clustering variable. Of the 108 patients fulfilling the inclusion criteria, 70 (64.8%) were discharged to rehabilitation. The variables independently associated with discharge to rehabilitation were younger age [odds ratio (OR)=0.89, 95% confidence interval (CI)=0.83-0.95, P=0.001], independent premorbid functional status (OR=14.92, 95% CI=2.43-91.60, P=0.004) and higher level of current mobility (OR=1.31, 95% CI=1.02-1.66, P less than 0.03). The multilevel model estimated that 12% of the total variability in discharge destination was explained by differences between the hospital units (p=0.12, 95% CI=0.02-0.55, P=0.048). The results indicate that the variables associated with discharge to rehabilitation following severe stroke are younger age, independent prestroke functional status and higher level of current mobility. In addition, organizational factors play a role in selection for rehabilitation, suggesting inequity in access for this patient group. (Contains 3 tables and 1 figure.)