Evaluated the relevance of the physiological-arousal model, the stimulus-control paradigm, and depression to insomnia both as a unitary construct and to its components. The Manifest Anxiety Scale, the Sleep Behavior Self Rating Scale and the Zung Depression Scale were administered to 81 clinical Ss. Three separate discriminant function analyses were performed with self-reported "sleeping difficulty," "latency of falling asleep," and "total hours of sleep" as criterion variables. The above three scales and the following four sleep patterns were used as indices: number of nocturnal wakings, latency to fall asleep once awake, number of early wakings, and frequency of feelings of fatigue upon wakings. The findings indicated that the physiological-arousal model was relevant both to insomnia overall and to its component of sleep-onset insomnia. The stimulus-control paradigm was found to be relevant only to sleep-onset insomnia. Depression was not a sensitive discriminator, possibly due to the heterogeneity of the patient population studied. It is emphasized that different mechanisms might be operating with the heterogeneous symptom "insomnia," and the replication of findings with criteria that include significant others and electroencephalographic measures is suggested.