Improving animal's temperament defined as "the animal's reaction to handling by human" could be a way to improve animal welfare. This study aimed at finding a simple measure for calf temperament discrimination which can be useful as a European selection criterion for on-farm beef cattle breeding schemes. Calves' reactions to human in the scale have been shown to be consistent across time, observer and with the docility test used to evaluate Limousin bulls in the French test stations. Therefore, we chose to test these measures on-farm for the purpose of this study. In order to ensure good family structure and connectedness across farms, 12 bulls were chosen among the Limousin bulls selected for artificial insemination use and their semen was diffused in 24 farms (56±23 weaned calves per sire). Behavioural records were registered at, on average, age of 218 ± 60 days for 1441 calves, offspring of 128 sires. For every calf, the quantity of rush movements (RM) and the total number of movements (TM) were scored using a continuous scale successively during weighing (P1), then when exposed to a stationary human situated in front of the scale for 10s (P2). The calf's reaction during the overall period was also assessed using the Australian crush test score on a 5-point scale ranging from 1 (docile) to 5 (aggressive). Mixed linear models were fitted to the data and variance components were estimated using Restricted Maximum Likelihood methods applied to linear animal models. Heritabilities were estimated at 0.28±0.09, 0.31±0.1, 0.19±0.07, 0.17±0.07 and 0.18±0.08 respectively for RM and TM scores measured during P1, RM and TM scores measured during P2 and the Australian score. All traits were highly genetically correlated (from 0.73±0.16 to 0.99±0.06). These results open the door to on-farm selection on docility in European conditions.