To test the repeatability of information about low-back pain elicited by self-administered questionnaire, histories obtained from 225 men and women were compared at an interval of 12 months. There was good agreement on whether subjects had ever suffered low-back pain (K = 0.82) and on whether the pain had ever led to consultation with a general practitioner (K = 0.76) or absence from work (K = 0.76). Information about the speed of onset of symptoms as well as histories of associated sciatica and disability for everyday activities were less reproducible. Epidemiologic studies based on such data must be interpreted with appropriate caution.