Introduction We tested the hypothesis that REM sleep contributes to core features of cognitive dysfunction of anxious depression including negative self-appraisals, biased memory processing and unpleasant dream content. Methods After a habituation night in a sleep lab, a convenience sample of 35 healthy college students and 20 depressed/anxious students were awakened 10 min into a REM sleep episode and then 10 min into a NREM sleep episode. Awakenings were counterbalanced to control circadian effects. After each awakening participants reported a dream and then completed memory recall, mood and self-appraisal tasks. Results Self-appraisals of depressed/anxious participants were significantly less positive and significantly more negative after awakenings from REM sleep vs NREM sleep. Appraisal of the REM sleep dream self was negative for depressed/anxious subjects only. Recall of negative memories was significantly more frequent after REM vs NREM sleep awakenings for both depress/anxious and healthy participants. REM sleep dreams were associated with greater frequencies of negative emotion, greater aggression and victimization rates than dreams in NREM sleep for depressed/anxious participants. Limitations Depressed/anxious participants were classified as such on the basis of mood scales rather than clinical interview. All participants were drawn from a volunteer college student population and thus our results may not be applicable to some elderly clinical populations. Conclusions REM appears to facilitate cognitive distortions of anxious depression.