Fifty-one healthy elderly subjects (median age = 65) gave retrospective estimates of nightmare frequency in questionnaires and recorded the occurrence of nightmares in daily logs over a 2-week period. (a) Mean annual nightmare frequency as estimated from logs was only 65% as high among college student controls. (b) Elderly subjects were about 1/5 as likely as college students to report a problem with nightmares. (c) Frequency estimates on the basis of logs were over 10 times higher than retrospective estimates.