Abstract: In an experiment on the effects of social environment and training on the human-animal relationship, 20 horses were handled according to a defined schedule. Eight horses were housed singly and 12 horses were housed in four groups of 3 horses. Horses were handled three times per week in 10 min sessions from an age of 6 months until 2 years of age during two winter periods. A total of 50 and 70 sessions were given in the first and second period, respectively. Five randomly allocated people performed the training. The training scheme involved leading, tying up, touching, lifting feet, etc. in 43 stages. The horse had to fulfil the performance criteria of each stage in order to get to the next stage. In the first winter period, horses were led to the stable when they had “passed” a stage or after 10 min of training. In the second winter period, horses would start off at stage 1 again, and when they “passed” a stage they went on to the next stage within the same training session. Because of the change in training procedure results were analysed separately for the two winter periods. There was a significant difference between trainers in the number of times they allowed a horse to “pass” a stage within each winter period (χ32, P<0.05; χ32, P<0.001 for the first and the second winter period, respectively). Group housed horses “passed” more stages than single housed horses (17 versus 14; 27 versus 18 in the first and second winter period, respectively; P<0.05 for the interaction). Singly housed horses bit the trainer more frequently than did group housed horses (P<0.01). The responses of group housed horses to training clearly demonstrate the benefits of raising young horses in groups.