Introduction Traumatic brain injury (TBI) is a commonly occurring condition and a significant public health issue, as it is one of the leading causes of disability worldwide. Complaints of disturbed sleep, excessive daytime sleepiness and disorders of arousal have been well established as being among the most pervasive, enduring and common sequelae, and have been shown to compromise the recovery process if left untreated. Current evidence suggests that sleep wake disorders begin in the acute stage with altered rest-activity cycles as evidenced from actigraphy findings. For those in the chronic stage, studies involving polysomnography have identified a number of changes in sleep architecture including reductions in total sleep time, sleep efficiency, and N3 sleep, and increased stage N1 sleep. For those with severe TBI, hypersomnia and increased sleep need have been identified as being the most common sleep complaint. To date however, there has not yet been an examination of sleep architecture in the acute stage of TBI. The objective of this study was to examine the sleep architecture of adults with acute moderate-severe traumatic brain injury, and to compare these findings with the sleep architecture of age- and sex-matched healthy controls. Materials and methods Six adults (4 males; 25±11.3 yrs), with moderate-severe TBI underwent 24-h polysomnography (PSG) on average 21days (range 7–38days) during their acute hospital stay. Results were compared with those of 11 healthy controls (7 males; 25±10.5yrs); who underwent in-laboratory PSG. Groups were compared for sleep architecture variables for the night period, determined as the longest ‘visually observable consecutive sleep period’ identified on the PSG recording in TBI patients. A Mann–Whitney test for non-parametric data was carried out to in order to compare PSG parameters between TBI patients and controls. Results Our results indicated that the TBI group (TBI) tended to have longer total sleep duration (TBI: 8.1±2.1h; Controls: 6.6±1.7h p=0.12), greater total wake duration (TBI: 159±104.7min; Controls: 53.6±47min, p=0.04), higher number of arousals and micro-arousals per hour of sleep (p=0.009), and greater number of sleep cycles (p=0.02). No group difference was found for the percentage of each sleep stage (N1: TBI: 46.1±11.8 %, Controls 42.6±16.2%; N2; TBI: 20.6±11.4%, Controls 28.5±11.5; N3; TBI: 29.8±17.0%, Controls: 28.3±11.0; REM: TBI: 17.2±6.0%, Controls 16.8±4.6%). Conclusion Although our sample size was small, these preliminary findings indicate that even in this very early stage, hospitalized patients with acute severe TBI experience a longer sleep duration, however with greater sleep fragmentation in comparison to control subjects. Additional research is needed to further elucidate and determine the course of these disturbances, and their relationship with clinical, neurological and functional outcomes. Acknowledgements We gratefully acknowledge the assistance of Ms. Hélène Blais for her assistance with the data for this project.