To determine how accurately elderly subjects recall recent falls, we studied 304 ambulatory men and women over the age of 60 years who completed a 12-month prospective study of risk factors for falling. We developed a system of weekly follow-up and home visits to record and confirm all falls. During the study, 179 participants suffered at least one fall that was confirmed by home visit. At the end of the study, all subjects were interviewed by telephone about whether they had suffered a fall during the preceding 3, 6, or 12 months. Depending on the time period of recall, 13% to 32% of those with confirmed falls did not recall falling during the specific period of time. Recall was better for the preceding 12 months than for 3 or 6 months. There were only weak correlations (r = 0.28 to 0.59) between the number of falls that were documented and the number that the subjects recalled during each of these periods. Those with lower scores on the Mini-Mental State Examination were more likely to forget falls. We conclude that elderly subjects often do not recall falls that occurred during specific periods of time over the preceding 3 to 12 months. Researchers and clinicians should consider using methods besides long-term recall for ascertaining and counting falls over specific periods of time.