Abstract: Horses kept in stalls are deprived of opportunities for social interactions, and the performance of natural behaviors is limited. Inadequate environmental conditions may compromise behavioral development. Initial training is a complex process and it is likely that the responses of horses may be affected by housing conditions. Sixteen 2-year-old Arabian horses were kept on pasture (P) (n=8) or in individual stalls (S) (n=8). Twelve horses (six P and six S) were subjected to a standardized training procedure, carried out by two trainers in a round pen, and 4 horses (two P and two S) were introduced to the round pen but were not trained (C; control). On sample collection day 0, 7, 21 and 28, behavior observations were carried out, blood samples were drawn and heart rates were monitored. Total training time for the stalled horses was significantly higher than total time for the pastured horses (S: 26.4+/-1.5 min; P: 19.7+/-1.1; P=0.032). The stalled group required more time to habituate to the activities occurring from the start of training to mounting (S: 11.4+/-0.96; P: 7.3+/-0.75 min; P=0.007). Frequency of unwanted behavior was higher in the stalled horses (S: 8.0+/-2.0; P: 2.2+/-1.0; P=0.020). Pastured horses tended to have higher basal heart rates on day 0 (S: 74.7+/-4.8; P: 81.8+/-5.3 bpm; P=0.0771). While the physiological data failed to identify differences between housing groups, the behavioral data suggest that pasture-kept horses adapt more easily to training than stalled horses.