Fifty-eight chronic nightmare sufferers were randomly assigned to two groups: treatment (n = 39) and wait-list control (n = 19). Treated Ss were taught a cognitive-behavioral technique called imagery rehearsal in which they learned in a waking state to change a nightmare and then to visualize the new set of images. Subjects were assessed pre-treatment and 3 months followup for nightmare frequency, self-rated distress and subjective sleep quality. Compared to controls, the treatment group showed significant and clinically meaningful decreases in nightmares. Treated Ss decreased nightmares as measured in nights/week (mean = -2.0, SD = 1.7, P = 0.0001) and actual number of nightmares (mean = -4.2, SD = 4.5, P = 0.0001). Significant improvement in self-rated sleep quality occurred in those treated compared with controls (P = 0.004); and, reduction in nightmares was a significant predictor of improvement in sleep (r = 0.55, P = 0.0001). These preliminary results lend support to the theory that, for some chronic sufferers, nightmares may be conceptualized as a primary sleep disorder which can be effectively and inexpensively treated with cognitive-behavioral therapy.