OBJECTIVE To measure the impact of Medicare's "75% rule" on readmission and death rates in elderly patients affected by the rule. DESIGN Retrospective study of two cohorts, both aged > or =65, discharged from a single medical center, from acute care with diagnoses excluded by the 75% rule. Group 1 (n = 4107) represented discharges in the year before the rule's enforcement and group 2 (n = 3893) for the rule's inaugural year. Logistic regression was used to compare mortality and readmission rates, and Cox regression was used for time to event data. RESULTS Overall, patients were readmitted and died relatively sooner in group 2. Mortality and readmission rates were significantly associated with an age and group interaction, with higher rates in group 2 among older patients. The increase in readmissions was greatest for pain syndromes (from 33 to 55%). In patients older than 85 with orthopedic diagnoses, the mortality rate increased from 25 to 54%. Cardiac patients died and were readmitted sooner in group 2 and pulmonary patients also died sooner. The largest subgroups, miscellaneous and lower limb joint replacement/osteoarthritis, did not show significant differences in readmission rates and mortality. CONCLUSIONS Restricting access to inpatient rehabilitation on the basis of diagnosis alone is associated with increased readmission and mortality, particularly in the very old. Comprehensive, evidence-based guidelines are needed to allocate rehabilitation services to those who need them most.