The central nervous pathways of angina pectoris have never been identified in vivo in man. We used positron emission tomography to examine the changes in regional cerebral blood flow associated with angina pectoris. Dynamic positron emission tomography with 15O-labelled water was used in 12 patients with angina and angiographically proven coronary artery disease to measure regional cerebral blood flow changes during angina induced by intravenous dobutamine. All subjects had typical retrosternal chest pain accompanied by ischaemic electrocardiographic changes during dobutamine infusion. Compared to the resting state, angina was associated with increased regional cerebral blood flow in the hypothalamus (percentage change in regional cerebral blood flow +6.5 and Z score 7.2) periaquaductal grey (+2.6 and 4.0), bilaterally in the thalamus (left: +2.7 and 4.3; right +3.7 and 4.7) and lateral prefrontal cortex (left +11.5 and 7.6; right +8.5 and 7.8) and left inferior anterocaudal cingulate cortex (+9.4 and 6.6). In contrast, it was reduced bilaterally in the mid-rostrocaudal cingulate cortex (left -3.7 and 6.3; right -4.7 and 4.6) and fusiform gyrus (left -3.2 and 4.0; right -3.3 and 3.7), right posterior cingulate (-3.9 and 5.8) and left parietal cortices (-4.8 and 6.3). Several minutes after stopping dobutamine infusion, when the patients no longer experienced angina and the electrocardiographic changes had resolved, thalamic, but not cortical activation could be seen. We propose that the central structures activated constitute the pathways for perception of anginal pain and that the thalamus may act as a gate to afferent pain signals, with cortical activation being necessary for the sensation of pain. This method of investigation may form a basis for research into anomalies of visceral pain perception such as silent myocardial ischaemia.