The objectives of this work were: to estimate the incidence of falls within an at-risk group of community-dwelling elderly people; to assess the risk factors associated with incident falls; to examine the effects of incident falls on survival. A random sample of 1042 community-dwelling older people in Nottingham (UK) were interviewed in 1985 and survivors re-interviewed at 4-year follow-up. The at-risk group was defined as survivors who had not fallen in the year prior to the baseline interview (n = 444). One-year fall recall was assessed using a questionnaire and included physical health, mobility, prescribed drugs and time spent walking. Body-weight and handgrip strength were measured. Eight-year post-fall mortality was recorded. In 1989 117 new fallers were identified. These people fell a total of 233 times in the year prior to re-interview (incidence rate: 524.8 per 1000 person-years at risk; 95% confidence intervals (CI): 473.3-576.3). People aged less than 75 were more likely to fall outdoors than people aged 75 and over (chi2 = 5.715, df = 1, p = 0.017). Risk factors associated with falling were: being less healthy (odds ratio (OR): 0.55; p = 0.052); having a walking speed in the range stroll/very slow/nonambulant compared with normal/brisk/fast (OR: 1.99; p<0.01); and number of prescribed drugs (OR: 1.30; p = 0.01). When analysed separately, indoor and outdoor falls presented differential risk profiles, with evidence that indoor falls were associated with frailty, while outdoor falls were associated with compromised health status in more active people. In 8-year post-fall monitoring, multiple (3+) fallers and indoor fallers showed a significant excess mortality. The differences in risk factors for, and prognoses following, indoor and outdoor falls, emphasise the complex interactions between intrinsic and extrinsic factors associated with falling among older people.