Despite the common misconception that rapid eye movement (REM) sleep is a unique correlate of dreaming, reports of mental activity can be elicited after awakenings from any stage of nocturnal sleep. We extended the investigation to naps and tried to explore the relationship between recall length and level of sleep stage and depth preceding the awakening. We hypothesized that dream report length would be related to arousal level. In 10 healthy young adults, sleep EEG and EOG were recorded for four non-consecutive early afternoon naps. Dream recalls were recorded following 10 s, 1 min, and 6 min of NREM Stage 2 and after 5 min of first REM period. We measured mental recall with total word count (TWC) method, sleep stages by using EEG visual scoring and Delta and Beta activity by period amplitude (PAA) and power spectral (PSA) analyses. All awakening conditions were followed by a dream report. TWC was significantly greater after REM than after 10 s and 1 min of NREM, and TWC did not differ among the NREM awakenings. Delta activity after REM was significantly lower compared to the NREM 6 and 1 min while Beta activity did not differ across the conditions. Assuming that arousal level decreased with increased NREM duration and increasing Delta EEG activity, the constant TWC across the three NREM awakenings indicates that arousal level cannot be the only factor affecting dream report length. Some other factor such as memory processing may explain the longer dream reports following REM sleep, or it may be that the EEG is an imperfect indicator of arousal level.