There is concern that interventions that use physical activity to prevent obesity in children might be undermined by an 'Activitystat', which exerts an effect to maintain a low set point for physical activity. The present critique summarises evidence from systematic reviews of interventions, from empirical tests of the Activitystat hypothesis, from studies on the heritability of physical activity in childhood and the physical activity of children of and adolescents across a wide range of physical and cultural environments. This body of evidence is inconsistent with the Activitystat hypothesis in its current form, and suggests that the emphasis on physical activity in obesity prevention interventions in children should be increased, not reduced.