This article is devoted to a consideration of the clinical prognostic significance of "tracé discontinu". The authors first distinguish "tracé discontinu" from the "tracé paroxystique" and tracé alternant" patterns in the newborn. The distinction is very important. The bad prognosis of the "tracé paroxystique" is well known, but prognosis of the "tracé discontinu" seems to be different. If the "tracé discontinu" is permanent, there was a 50% favourable outcome in our 22 newborn babies brought to the reanimation department. If some continuous activity can be observed: beginning of sleep organisation, then the prognosis seems to be better: 61% of our cases. As blood levels of anticonvulsants are rarely measured it seems difficult, given our present knowledge, to appreciate anticonvulsive drug effects on EEG recordings. Thus 3 aims should be pointed out: --Measurements of anticonvulsant blood level each time a "tracé discontinu" is observed in a treated newborn. --Early EEG recordings of sufficient duration to obtain eventually some continuous tracing. --Correct differentiation between "tracé paroxystique" and "tracé discontinu" for the clinician: "tracé discontinu" in a full-term newborn should never allow the paediatrician to stop treatment.