To evaluate the effects of environmental enrichment on behavioral, physiological and productive characteristics, 71 Japanese Black x Holstein steers (8 months of age; 299.5 +/- 22.6 kg) were allocated to three pens in two repetitive experiments. Pen C (n = 11 and 12) consisted of a feeding alley for grain feed, a trough for dry hay, a water bowl and a resting space as a control pen. Pen D (n = 12 and 12) included a drum can (58 cm diameter x 90 cm height) containing hay. Pen GD (n = 12 and 12) included a drum can that was placed around artificial plastic turf (30 x 120 cm) for grooming. The drum cans were removed after 5 months of installation. Behavioral observations were made for 2 h at 10 min intervals after feeding on three successive days each month for 10 months. Agonistic interactions were also continuously observed for 1 h after feeding to assess the dominance order (DO). Sampling blood and measuring bodyweight were performed bimonthly. The steers used the drum can frequently for 3 months after installation (1st, 2nd, 3rd months vs 4 months, all P < 0.05). The frequency of total eating of grain feed and hay was higher in pen D and pen GD than in pen C (both P < 0.01), while it was lowest in pen GD after removal of the drum can (P < 0.05). Grooming at the drum can was observed more frequently in pen GD than in pen D (P < 0.05). After they finished eating the grain feed, they ate hay at the drum can that contained additional hay rather than at the trough for hay (P < 0.01). Plasma dopamine concentrations were higher in pen D than in pen C (P < 0.05), and serum triglyceride concentrations were higher in pen C than in pen GD (P < 0.05) during the installation of the drum can. After removal of the drum can, serum total cholesterol concentrations became higher in pen D and GD than in pen C (both P < 0.05). Average daily gain correlated positively with the frequency of eating hay at the drum can in pen D (r(s) = 0.52, P < 0.01). In pen GD, the frequency of using the drum can correlated negatively with DO (r(s)=-0.59, P < 0.01). Carcass belly fat was thicker in pens D and GD than in pen C (both P < 0.01). In pen GD, the frequency of eating hay (r(s) = 0.79, P < 0.01) and grooming at the drum can (r(s) = 0.63, P < 0.05) correlated positively with the marbling score. Although social factor affected the steers using the drum can, installing it in the early fattening stage encouraged the steers to eat and groom there and resulted in better carcass characteristics through the prolonged physiological positive effects.