Eighty-one elderly residents of a hostel for the aged (mean age 83.3 years) underwent clinical medical assessments to examine susceptibility to falling. The medical examination was structured and followed a clinical format with particular emphasis upon posture and gait, in addition to routine examination of the cardiovascular, respiratory, neurological, gastrointestinal, haemopoietic, genitourinary, musculoskeletal and visual systems. These subjects were then followed up for 1 year to assess whether these clinical measures were associated with falls. Seventy-six residents were available for follow-up. Thirty-four subjects (44.7%) reported having no falls in the follow-up year, 10 (13.2%) fell once only, 13 (17.1%) fell on two occasions whilst 19 (25.0%) fell three or more times. Seventy-two percent of all falls occurred in the hostel building. Certain clinical factors showed high specificity (i.e. only a few non-fallers screened positive in the tests) but low sensitivity (i.e. there were considerable numbers of residents who screened negatively, but fell in the 12-month follow-up period). Stepwise logistic regression analysis revealed impaired cognition, abnormal reaction to any push or pressure, history of palpitations and abnormal stepping as variables that independently and significantly predicted falling. The equation predicted falls with 70.7% sensitivity and 79.4% specificity, with an overall predictive accuracy of 74.7%. It appears that a modified focussed clinical examination could provide the basis of a short assessment for predicting falls and highlight possible intervention strategies for reducing fall risk.