Abstract Background Information is scarce on unplanned transfers from geriatric rehabilitation back to acute care despite their potential impact on patients’ functional recovery. This study aimed 1) to determine the incidence rate and causes of unplanned transfers; 2) to compare the characteristics and outcomes of patients with and without unplanned transfer. Methods Consecutive stays (n = 2375) in a tertiary geriatric rehabilitation unit were included. Unplanned transfers to acute care and their causes were analyzed from discharge summaries. Data on patients’ socio-demographics, health, functional, and mental status; length of stay; discharge destination; and death, were extracted from the hospital database. Bi- and multi-variable analyses investigated the association between patients’ characteristics and unplanned transfers. Results One in six (16.7%) rehabilitation stays was interrupted by a transfer, most often secondary to infections (19.3%), cardiac (16.8%), abdominal (12.7%), trauma (12.2%), and neurological problems (9.4%). Older patients (AdjORage≥85: 0.70; 95%CI: 0. 53–0.94, P = .016), and those admitted for gait disorders (AdjOR: 0.73; 95%CI: 0.53–0.99, P = .046) had lower odds of transfer to acute care. In contrast, men (AdjOR: 1.71; 95%CI: 1.29–2.26, P < .001), patients with more severe disease (AdjORCIRS: 1.05; 95%CI: 1.02–1.07, P < .001), functional impairment before (AdjOR: 1.69; 95%CI: 1.05–2.70, P = .029) and at rehabilitation admission (AdjOR: 2.07; 95%CI: 1.56- 2.76, P < .001) had higher odds of transfer. Transferred patients were significantly more likely to die than those without transfer (AdjOR 13.78; 95%CI: 6.46–29.42, P < .001) during their stay, but those surviving had similar functional performance and rate of home discharge at the end of the stay. Conclusion A significant minority of patients experienced an unplanned transfer that potentially interfered with their rehabilitation and was associated with poorer outcomes. Men, patients with more severe disease and functional impairment appear at increased risk. Further studies should investigate whether interventions targeting these patients may prevent unplanned transfers and modify associated adverse outcomes.