Adequate sleep is important for the health and well being of animals but we lack non-invasive methods to record sleep states from group-housed, freely-moving farm animals. We used electrophysiological data (electroencephalography; EEG, electromyography; EMG, and electro-oculography; EOG) to characterize sleep states in calves and examined how well observations of resting behaviour were correlated with the electrophysiological data. We obtained 20h of EEG, EMG and EOG recordings from each of six pair-housed dairy calves using an ambulatory EEG recording device, while recording their resting posture via direct observation. Visual scoring of the electrophysiological data was used to distinguish between awake, rapid eye movement (REM) sleep, and non-rapid eye movement (NREM) sleep during 30s epochs. Calves were asleep for a mean (±S.D.) 25% (±2.0) of all observations in 50 (±22) bouts of 5 (±2) min per day. NREM sleep composed 55% (±7) and REM sleep 45% (±7) of the calves’ total sleep time. Both types of sleep occurred in short bouts of 2–3min. According to electrophysiological data, calves were awake during 81% (±9) of the 30s-epochs when the calf was standing or resting with its head lifted up and moving. Observations of “the calf resting head lifted up still” predicted 55% (±9) of the epochs of NREM sleep. The best behavioural predictor of REM sleep was “the calf resting with neck relaxed”, which predicted 61% (±9) of the epochs of REMS. The calves’ resting body postures can successfully be used to estimate calves’ total duration of time asleep and duration of time spent in REM and NREM sleep states. However, there was less success in estimating the time spent in the different phases of sleep during each 30s epoch. Electrophysiological data can be recorded non-invasively from freely-moving, group-housed calves and observations of resting behaviour can identify when calves are asleep, although further work is needed to use behaviour to identify the phases of sleep.