Our neurobiological knowledge about human dream organisation results primarily from the study of Rapid-Eye-Movement (REM) sleep. In humans, functional neuroimaging techniques, using H215O or 18FDG positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), allowed the mapping of the regional cerebral activity during this sleep stage, which is dominated by the activation of the pons, the thalamus, temporo-occipital and limbic/paralimbic areas (including the amygdala, the hippocampal formation and the anterior cingulate cortex), along with a relative quiescence of dorsolateral prefrontal and inferior parietal cortices. These results are in agreement with animal neurophysiological data about REM-sleep generation. They may also explain several hallmarks of dreaming experience that are found in dream reports after awakening from REM sleep. For instance, amygdala activation is consistent with the predominance of threat-related emotions. Temporo-occipital activation is in keeping with visual dream imagery. Prefrontal deactivation is suggestive of the lack of orientational stability, the alteration in time perception, the delusional belief of being awake, the decrease in volitional control and the fragmented episodic memory recall. Inferior parietal deactivation may contribute to the lack of distinction between first- and third-person perspectives. Conversely, specific cognitive and emotional features in individual dreams could be used to predict some aspects of the regional functional organisation of the human brain during dreams, and also inspire the design of future dedicated neuroimaging studies by offering constraints to the analysis and interpretation of sleep data acquired just before dream reports. Therefore, we suggest that future functional brain imaging in humans should be combined with a careful neuropsychological analysis of dream reports, and especially their categorisation based on the presence of specific bizarre features, to test hypotheses about the brain correlates of dreams. As little is known about the physiology of non-REM sleep dreaming, future neuroimaging studies should also attempt to link dreaming experiences during this sleep stage with patterns of regional cerebral activity. Overall, although many questions arising from the study of oneiric behaviour remain unanswered, recent neurophysiological and neuroimaging research about REM sleep offers an increasingly detailed picture of the cerebral correlates of dreaming that might also provide new insights into dream functions.