The objective of the study was to examine the predictive effect of prospectively registered falls on survival within a randomly selected group of elderly women. A longitudinal study, with 9 years follows up after 1 year prospective fall registration was designed. Persons age 75 or more living in the community. A total of 300 Norwegian women; mean age was 80.8 participated in the study. Base-line registrations of health, functioning and medication were made. Falls which occurred during a 1-year period after baseline were registered as well as the date of death within a 9-year period after the end of the fall registration period. Half of the women reported one or more falls (1-11). Eighty-six women (28.7%) experienced only one fall and 65 (21.7%) had at least two. During the 9-year follow-up period, 41.7% died. Cox regression analyzes demonstrated that frequent falling, old age and a self-reported worsening of health were significantly associated with mortality during the follow-up period. Their relative risks of death when experiencing at least two falls was 1.6 (95% CI 1.1-2.4), P = 0.04, when compared with no falls. Older fallers appear to have markedly increased mortality. Since falls are common among elderly people, this is a relevant fact for public health policy. Increasing age, poor self-rated health and high frequency of falls predict independently mortality in our 9 years follow up study. Because women represent the largest proportion of the elderly and falls are amendable, fall preventive may have the potential for significant impact on increasing year of life.