A major problem confronting behavioural theories of hypertension, such as the reactivity hypothesis, is that stress is likely to be intermittent, whereas the early stages of hypertension appear to be characterized by an increase in the tonic level of blood pressure and sympathetic activity. Furthermore, intermittent sympathetic arousal (e.g. exercise, thigh-cuff compression) does not necessarily raise tonic blood pressure. A worksite-based study of occupational stress has indicated that people in high-stress jobs have increased blood pressure throughout the day and night, which is at least consistent with a behaviourally mediated resetting of the tonic blood pressure level. There is evidence that adrenaline is preferentially released in response to behavioural stresses. According to the 'adrenaline hypothesis', adrenaline can raise tonic blood pressure while noradrenaline does not. We therefore propose that the different long-term effects of behavioural stress and exercise on blood pressure can be explained by their differing effects on catecholamine release.