The electroencephalogram (EEG), an easy-to-use and non invasive cerebral investigation, is a useful tool for diagnosis and early prognosis in newborn babies. In newborn full term babies manifesting abnormal clinical signs, EEG can point focal lesions or specific aetiology. EEG background activity and sleep organization have a high prognostic value. Tracings recorded over long period can detect seizures, with or without clinical manifestations, and differentiate them from paroxysmal non epileptic movements. The EEG should therefore be recorded at the beginning of the first symptoms, and if possible before any seizure treatment. When used as a neonatal prognostic tool, EEG background activity is classified as normal, abnormal (type A and type B discontinuous and hyperactive rapid tracing) or highly abnormal (inactive, paroxysmal, low voltage plus theta tracing). In such cases, the initial recording must be made between 12 and 48 h after birth, and then between 4 and 8 days of life. Severe EEG abnormalities before 12 h of life have no reliable prognostic value but may help in the choice of early neuroprotective treatment of acute cerebral hypoxia-ischemia. During presumed hypoxic-ischemic encephalopathy, unusual EEG patterns may indicate another diagnosis. In premature newborn babies (29-32 w GA) with neurological abnormalities, EEG use is the same as in term newborns. Without any neurological abnormal sign, EEG requirements depend on GA and the mother's or child's risk factors. Before 28 w GA, when looking for positive rolandic sharp waves (PRSW), EEG records are to be acquired systematically at D2-D3, D7-D8, 31-32 and 36 w GA. It is well known that numerous and persistent PRSW are related to periventricular leukomalacia (PVL) and indicate a bad prognosis. In babies born after 32 GA with clinically severe symptoms, an EEG should be performed before D7. Background activity, organization and maturation of the tracing are valuable diagnosis and prognosis indicators. These recommendations are designed (1) to get a maximum of precise informations from a limited number of tracings and (2) to standardize practices and thus facilitate comparisons and multicenter studies.