The validity and reliability of patient recall of diseases depends on the kind of disease. The validity and reliability of patients' recall of fractures was assessed among the 146 cases of a nested case-control study on independent living following fractures in the elderly. On declaration of a fracture a letter was sent to the general practitioner requesting medical information on the latest fracture receiving treatment. A total of 99 medical discharge letters (68%) could be collected for further evaluation. The information on fracture localisation which was obtained both during the patients interviews 1994/95 and 1998 and through the medical discharge letters were coded with reference to the AO (Arbeitsgemeinschaft für Osteosynthesefragen)-classification and compared. The Kappa-statistic for the reliability of patient information was 0.80-0.89, and 0.77-0.89 for the validity, depending on the degree of specified concordance. Obtaining information on fractures from patients appears as a valid and reliable source of epidemiologic data. Patients who generally are affected physically or had to stay in a hospital due to a fracture or its treatment seem to have a good recall of this event. It appears that it can be remembered years after the event with sufficient reliability and validity.