This study investigated the consistency and stability of fear responses in housed beef heifers over time using three behavioural tests. The flight (time to join peers and avoidance distance), docility (isolation and handling) and fear (4 phases; responses of isolated heifers in (i), the absence (ii), the presence, of food and responses to a stationary human (iii) without and (iv) with visual contact with their peers) tests were carried out over three consecutive days, in that order, commencing on day 30 and again on day 81 post-housing. Animals were blood sampled twice, in a chute, between the two test sessions. Scores (low scores = more fearful) were assigned to the fear responses. Ten out of 19 variables were consistent over time. Consistency was more evident in the presence of a human, whereas agitation and avoidance of the animals were the most reliable measures. For each session, Principal Component Analysis showed that these behaviours were also consistent within different situations and that the two first components explained more than 55% of the variation of the data. Crossbred heifers showed stability for the consistent behaviours while purebred heifers were more fearful during the second test session. This study showed that fearfulness is a multi dimensional construct and that genetic factors as well as environmental factors affected the behavioural responses. Human–animal relationship can be assessed by measuring agitation while general fearfulness requires tests measuring different dimensions.