In this study we test the hypothesis that selecting sheep for a low behavioural reactivity to stressful situations will improve their metabolic efficiency, and thereby feed efficiency, during a controlled trial in an animal house. Twenty-four Merino wethers were used, 12 each from lines selected for high (HBR) and low (LBR) behavioural reactivity to stressful stimuli (human presence and social isolation). The sheep were habituated to the experimental procedures for 10 days, followed by 45 days during which voluntary feed intake was measured so that total daily energy intake was quantified. The sheep were weighed twice weekly before daily feeding. Feed efficiency was determined by measuring net feed intake, average daily weight gain and body condition score. Our hypothesis was not supported by the results of this study. There was no difference between LBR and HBR sheep in average daily weight gain or body condition score. The net feed intake of HBR sheep was lower than that of LBR sheep (P = 0.02), indicating that under the conditions of our experiment, HBR sheep were actually more feed efficient than LBR sheep. This study was carried out on sheep with steady intakes and in familiar surroundings. It is possible that LBR sheep may be more efficient than HBR sheep in more stressful situations.