We present a comprehensive review of sleep studies performed in patients with brain lesions complemented by 16 additional personal selected cases and by discussion of the corresponding animal data. The reader is cautioned about the risk of establishing an erroneous correlation between abnormal sleep and a given disorder due to the important inter and intra variability of sleep parameters among individuals. Salient points are stressed: the high frequency of post-stroke sleep breathing disorders is becoming increasingly recognised and may, in the near future, change the way this condition is managed. Meso-diencephalic bilateral infarcts induce a variable degree of damage to both waking and non-REM sleep networks producing and abnormal waking and sometimes a stage 1 hypersomnia reduced by modafinil or bromocriptine, which can be considered as a syndrome of cathecholaminergic deficiency. Central pontine lesions induce REM and non-REM sleep insomnia with bilateral lateral gaze paralysis. Bulbar stroke leads to frequent sleep breathing disorders. Polysomnography can help define the extent of involvement of various degenerative diseases. Fragmented sleep in Parkinson's disease may be preceded by REM sleep behavioural disorders. Multiple system atrophies are characterised by important sleep disorganization. Sleep waking disorganization and a specific ocular REM pattern are often seen in supra-nuclear ophtalmoplegia. In Alzheimer patients, sleep perturbations parallel the mental deterioration and are possibly related to cholinergic deficiency. Fronto-temporal dementia may be associated with an important decrease in REM sleep. Few narcoleptic syndromes are reported to be associated with a tumour of the third ventricle or a multiple sclerosis or to follow a brain trauma; all these cases raise the question whether this is a simple coincidence, a revelation of a latent narcolepsy or, as in non-DR16/DQ5 patients, a genuine symptomatic narcolepsy. Trypanosomiasis and the abnormal prion protein precociously after sleep patterns. Polysomnography is a precious tool for evaluating brain function provided it is realised under optimal conditions in stable patients and interpreted with caution. Several unpublished cases are presented: one case of pseudohypersomnia due to a bilateral thalamic infarct and corrected by modafinil, four probable late-onset autosomal recessive cerebellar ataxias without sleep pattern anomalies, six cases of fronto-temporal dementia with strong reduction in total sleep time and REMS percentage on the first polysomnographic night, one case of periodic hypersomnia associated with a Rathke's cleft cyst and four cases of suspected symptomatic narcolepsy with a DR16-DQ5 haplotype, three of which were post-traumatic without MRI anomalies, and one associated with multiple sclerosis exhibiting pontine hyper signals on MRI.