Over the past few years, research on remote monitoring of animal behaviour by means of accelerometers integrated in GPS collars considerably increased. Use of accelerometers allows for long-term fine-scale behavioural measurements, which are extremely useful to study activity patterns. As the values generated by accelerometers are strongly affected by individual factors, season-related environmental effects, and the position of the collar on the animal, comparisons of accelerometer data among different individuals and time-periods may yield misleading results. Researchers have to find an easy-to-use method in order to turn accelerometer data into behavioural data, one which enables them to take into consideration inter- and intra-individual variations. We propose an easy individual-based method, which generates threshold values to distinguish between active and inactive behaviours with no need of direct observation. By treating each animal independently and adopting ad hoc temporal scales, this method is able to take into consideration the influence of individual factor modifications (e.g., body size, collar tightness) on the data recorded by the accelerometer. We validated this approach and showed its potential by testing it with an activity dataset from 26 free-ranging Alpine ibex (Capra ibex). We managed to distinguish between active and inactive behaviours with a high percentage (93%) of correctly classified binary behavioural state. We showed that, when the threshold values are calculated at a large temporal scale, the accuracy decreases and activity pattern predictions may yield misleading results. By adopting the method proposed and by transforming the accelerometer data provided by the collars into time spent active, it may be possible to analyse how the activity levels of the monitored individuals change over the seasons, to appreciate fine variations of individual characteristics, and to compare the activity patterns of different populations as well as of different species.