The significance of arousal in apnea termination in preterm infants is not known. We investigated the appearance of arousals from sleep with polysomnography for 21 preterm infants at a median age of 36 gestational weeks. The polysomnographic appearance of sleep was fragmented by frequent arousals. The number of spontaneous arousals unrelated to apneas was 18 per hour in sleep; higher in rapid eye movement (REM) sleep than in non-REM sleep (p < 0.001). Eighty-two percent of arousals were regarded as spontaneous, and 18% were related to apneas. In turn, arousal followed 5% of all apneas; 30% of mixed, 2% of central, and 20% of long apneas defined as apnea of prematurity. Apneas without an arousal led to lower oxygen saturation levels than those followed by an arousal (p < 0.001). Mixed apneas with an arousal had stronger breathing effort and a higher number of breaths compared with apneas without an arousal (p < 0.05). In preterm infants, frequent spontaneous arousals or arousal-type phenomena make the polysomnographic appearance of sleep fragmented. However, even long apneas or hypoxia commonly fail to elicit arousals or any sign of sleep interruption. Our findings suggest that arousal appears not to be the main mechanism for apnea termination in preterm infants. Polysomnographic appearance of sleep in preterm infants is fragmented by arousals. Contrary to older children and adults, arousal to apnea is uncommon in preterm infants. Even long mixed apneas with desaturation mostly fail to elicit an arousal response. In preterm infants, apnea termination appears not to depend on an arousal. Low arousability is suggested to be caused by a low ventilation response to hypoxia.