Technological innovations have made heat-sensitive data-loggers smaller, more efficient and less expensive, which has led to a growing body of literature that measures the skin- or body temperatures of small animals in their natural environments. Studies of this type on heterothermic endotherms have prompted much debate regarding how to best define ‘torpor’ expressions from skin- or body temperature data alone. We propose a new quantitative method for defining torpor ‘entries’, ‘arousals’ and ‘stable torpor periods’ whilst comparing the results to ‘torpor bout’ durations identified using only the torpor cut-off method. By decomposing a torpor bout into ‘entries’, ‘stable torpor periods’, and ‘active arousals’, we avoid biases introduced by using strict threshold temperature values for the onset of torpor, thereby allowing better insight into individual use of torpor. We present our method as an easy-to-use function written in R-code, offering an un-biased and consistent methodology to be applied on skin- or body temperature measurements across datasets and research groups. When testing the function on a large dataset of skin temperature data collected on three bat species in Norway (Plecotus auritus: Nind = 39; Eptesicus nilssonii: Nind = 11; Myotis brandtii: Nind = 10), we identified 461 complete torpor bouts across species. More than 40% of the torpor bouts (Nbouts = 192) did not contain stable torpor periods, because the bats aroused before they had reached a stable skin temperature level. Furthermore, only considering ‘torpid’ and ‘euthermic’ temperature values by applying strict cut-off thresholds led to potentially large underestimations of torpor bout durations compared to our quantitative determination of the onset and termination of each torpor bout. We highlight the importance of differentiating between torpor phases, especially for active arousals that can be very energetically expensive and may alter our evaluation of the actual energetic savings gained by an individual employing torpor.