Résumé : Preterm newborns are at high risk of neurological injury. In this population, we investigated the link between neurological complications and sleep architecture. At term-corrected gestational age, we studied retrospectively the polysomnography of 45 preterm infants born at < 28 weeks or weighting < 1 kg. These infants were followed-up by a neuropaediatrician (median age at last follow-up 50.4 months). Two groups of children were constituted: a group without neurological disorder and a second group with at least one of the following: cerebral palsy, language or mental retardation, visual or hearing disability or attention disorder. A Multiple Indicators and Multiple Causes model assessed the relationship between the neurological outcome and two sleep components: spontaneous arousability [number of awakenings and movements per hour of quiet sleep (QS) and active sleep] and QS characteristics (median duration of QS cycles and percentage of QS over total sleep time). Twenty-six infants had an impaired neurological outcome. There were no statistical differences between the two groups regarding clinical characteristics. Compared to preterm neonates with normal neurological outcome, those with impaired outcomes had a lower spontaneous arousability; i.e. 0.7 (0.5–1) times less awakenings and movements per hour of QS and 0.9 (0.8–1) times less per hour of active sleep than infants with normal outcomes (P = 0.05). The differences in QS characteristics did not reach statistical significance. These findings suggested that, in preterm infants, perinatal neurological injuries could be associated with an abnormal sleep architecture characterized by altered spontaneous arousability.