The effect of dietary carbohydrate source on physical activity in relation to metabolic rate in pigs was studied. Six groups of 12 pigs (50-kg, castrated males) were fed one of two diets: a starch diet or a non-starch polysaccharide (NSP) diet. Both diets had a similar calculated net energy content. The starch diet contained 13% tapioca, and the NSP diet contained 17% sugar beet pulp silage. Pigs were housed in groups and fed at approximately two times maintenance (approximately 900 kJ ME.kg-.75.d-1). Nitrogen and energy balances were measured per group during a 7-d experimental period, which was preceded by a 2-wk adaptation period. Heat production and physical activity were measured during successive discrete 9-min intervals. Metabolizability of gross energy was 79.4% and 78.2% for the starch-fed and NSP-fed pigs, respectively (P < .10). Activity differed between treatments. Activity-related heat production tended to be higher for the starch-fed pigs than for those fed the NSP diet (P < .10); 118 and 90 kJ.kg-.75.d-1, respectively. During the dark phase of the day (from 2000 to 0800) activity-related heat production was 8 kJ.kg-.75.d-1 higher for the starch fed pigs than for those fed the NSP diet (P > .1) whereas during the light phase, this difference was 48 kJ.kg-.75.d-1 (P < .10). Retained energy was similar for both treatments. These results show that dietary composition can change physical activity of pigs and thereby affect their energy utilization.