ABSTRACTOne difficulty associated with the study of dreams has been the absence of an objective indicator of dreaming in sleep other than REM. This study explores the use of a subjective judgment and instrumental response to mark the occurrence of dreams. Content reports of 8 subjects collected under two conditions were compared. On 2 nights the experimenter determined the time of the awakening and on the other 2 the subject's behavioral response was used as the indicator. The subjects' judgments that they were dreaming proved to be more accurate than was the experimenter's. A single case study of a high NREM responder carried out for 9 nights showed the frequency of NREM dreaming signals to covary with presleep mood.