Tracking devices are commonly used to locate and monitor wild animals for studying spatial ecology and survival rates. There is growing interest in capture effects, partially to minimize the impact on the study species, but also for animal welfare reasons. This study aims to examine roe deer behaviour in box-traps, when restrained, when released and during recaptures to quantify injuries and deaths over a period of 41 years. We use data from 2911 captures from 926 individuals between 1973 and 2014. We recorded behaviour inside the box-traps over two seasons. We also recorded behavioural data from 671 catches of 346 individuals during six seasons to study habituation. Additionally, we discuss box-traps in relation to ethological theory and animal welfare. Over a 41-year period, one roe deer suffering from starvation was found dead in a trap (0.035%), which cannot be solely related to capture (N = 926). About 58% of all roe deer were recaptured at least once during their life time. There was a low prevalence of injuries (0.5% of the captures, N = 2911), and they occurred predominately to the nose or antlers in velvet (in males). During the first hour after capture, animals typically stand very tense between eating bouts. Thereafter, the deer tended to move more softly and exhibited resting behaviours (e.g. lying down). Overall, we conclude that this method of capture and handling had very low impact on the welfare or survival of roe deer, which also habituated to recapture over successive events.