Objectives: To assess the effi ciency of a short-stay unit (SSU) for undiff erentiated medical patients and evaluate its eff ect on the overall efficiency of a general medicine department. Design, setting and patients: Retrospective study of all general medical patients admitted to the SSU at Flinders Medical Centre, South Australia, during its 5 years of operation (2005-2009), compared with 4 years before its institution and 2 years after its closure. Main outcome measures: Relative stay index (RSI); inhospital mortality; readmissions within 7 and 28 days. Results: 23 790 general medical patients were admitted overall, and 10 764 of these (45.2%) were admitted to the SSU. The RSI for the SSU during its years of operation was 0.79, compared with 1.34 for the long-stay unit. The overall RSI for the department did not improve during those years and was not signifi cantly diff erent to the periods before or after. Conclusions: We found no evidence that an SSU for undiff erentiated medical patients creates bed capacity. It does, however, appear to be safe.