Retrospective measurements underestimate nightmare frequency, but little is known about how retrospective
duration and attention for nightmares may affect this process. This study evaluates the differences between two retrospective
durations, a prospective log, and a retrospective estimate after keeping this log. Forty-nine participants completed the
SLEEP-50, kept a nightmare log, and estimated their nightmares after keeping a log. Paired t-tests showed that estimates
of nightmare frequencies differed significantly from each other according to measurement type (p < .05). Prospective logs
are the most accurate way of estimating nightmare frequency, possibly due to simply forgetting over time. For treatment
studies relying solely on retrospective measurements, a short duration is recommended.