Semi-aquatic mammals have evolved to forage in both aquatic and terrestrial environments, and they generally lack locomotor specialisation for either. Having relatively unspecialised adaptations, some semi-aquatic species are generalist foragers, and their activity and diving patterns provide insight into constraints on their foraging behaviour. The recent miniaturisation and improved accuracy of logging devices allow remote data collection from small (<1kg), shallow-diving species. The development of analytical methods has lagged behind technological advances (and techniques developed for fully aquatic animals do not account for the variability of behaviour typical of semi-aquatic animals and their switching between terrestrial and aquatic environments). I applied and developed novel analytical techniques to identify activities and diving patterns of a semi-aquatic mustelid, the American mink (Neovison vison), fitted with Time-Depth Recorders (TDRs). Using a hidden Markov model (HMM) algorithm allowed me to classify dives into three states to identify clustering, and describe sequential diving patterns by mink. TDRs were used to determine active/inactive periods on the basis of rapid TDR temperature changes, and this method was validated empirically. Having developed this methodology, I applied it to 18 datasets collected from 14 mink in lowland England. Terrestrial activity of mink was positively related to ambient temperature (across both sexes), however aquatic activity, especially diving, appeared to be more influenced by daylight than by temperature. Mink showed intersexual differences, with males being more nocturnal and more active on land, and females more diurnal, and more persistent in diving. There was considerable variability between sexes and individuals. This is the first study to use HMM to classify the dives of a semi-aquatic animal, and the first to use TDR temperature records to identify mammalian activity patterns. These methods will be generally applicable to animals that make rapid transitions between environments and have thus far been difficult to study.