Previous functional imaging studies have demonstrated a number of discrete brain structures that increase activity with noxious stimulation. Of the commonly identified central structures, only the anterior cingulate cortex shows a consistent response during the experience of pain. The insula and thalamus demonstrate reasonable consistency while all other regions, including the lentiform nucleus, somatosensory cortex and prefrontal cortex, are active in no more than half the current studies. The reason for such discrepancy is likely to be due in part to methodological variability and in part to individual variability. One aspect of the methodology which is likely to contribute is the stimulus intensity. Studies vary considerably regarding the intensity of the noxious and non-noxious stimuli delivered. This is likely to produce varying activation of central structures coding for the intensity, affective and cognitive components of pain. Using twelve healthy volunteers and positron emission tomography (PET), the regional cerebral blood flow (rCBF) responses to four intensities of stimulation were recorded. The stimulation was delivered by a CO2 laser and was described subjectively as either warm (not painful), pain threshold just painful), mildly painful or moderately painful. The following group subtractions were made to examine the changing cerebral responses as the stimulus intensity increased: (1) just painful - warm; (2) mild pain - warm; and (3) moderate pain - warm. In addition, rCBF changes were correlated with the subjective stimulus ratings. The results for comparison '1' indicated activity in the contralateral prefrontal (area 10/46/44), bilateral inferior parietal (area 40) and ipsilateral premotor cortices (area 6), possibly reflecting initial orientation and plans for movement. The latter comparisons and correlation analysis indicated a wide range of active regions including bilateral prefrontal, inferior parietal and premotor cortices and thalamic responses, contralateral hippocampus, insula and primary somatosensory cortex and ipsilateral perigenual cingulate cortex (area 24) and medial frontal cortex (area 32). Decreased rCBF was observed in the amygdala region. These responses were interpreted with respect to their contribution to the multidimensional aspects of pain including fear avoidance, affect, sensation and motivation or motor initiation. It is suggested that future studies examine the precise roles of each particular region during the central processing of pain.