Studies on nightmare frequency have yielded inconsistent results. We compared the frequency of nightmares and bad dreams obtained with retrospective methods (annual and monthly estimates) and with two types of prospective measures (narrative and checklist logs). Four hundred and eleven participants completed retrospective estimates of nightmare and bad dream frequency and recorded their dreams in either narrative or checklist logs for 2–5 weeks. When measured prospectively with narrative logs, nightmare frequency was marginally higher than the 1-year estimate (P= 0.057) but not significantly different from the 1-month estimate (P> 0.05). Prospective bad dream frequency was significantly greater than the two retrospective estimates (ps < 0.0005). There were no significant differences in the frequency of nightmares and bad dreams reported prospectively with narrative versus checklist logs (ps > 0.05). However, checklist logs yielded a significantly greater number of everyday dreams per week (P< 0.0001). Taken together, the results provide partial support for the idea that when compared to daily logs, retrospective self-reports significantly underestimate current nightmare and bad dream frequency. Prospective studies of dream recall and nightmare frequency should take into account the type of log used, its duration, and the participants' level of motivation over time.