This prospective study was designed to clarify the association between the incidence of falls and physical fitness in healthy community-dwelling elderly people without history of fall-related injuries within 1-5 years before the baseline assessment. Six hundred and one elderly subjects aged 60-79 years who lived independently participated in baseline assessments from March 2002 to March 2004, and were followed up for one year to detect falling. Demographic items (age and sex), chronic disease, self-rated health, difficulty in performing any activity, tendency to stay at home, body mass index (BMI), %vital capacity, handgrip and knee extension strength, stepping rate, one-leg standing with eyes opened, trunk flexion at sitting position, maximum walking speed, instrumental activity of daily living (IADL), and General Health Questionnaire (GHQ30) were assessed at baseline. Four hundred and eighty-one participants (80.0%) completed the 1-year follow-up.The fall incidence was 13.1%. In men, fallers had lower BMI and decreased maximum walking speed, however, neither showed significant odds ratio (OR) for fall. In women, fallers had lower knee extension strength and the ratio of weak side to strong side (WS/SS ratio) in knee extension strength compared to non-fallers. The age-adjusted OR for fallers in the lowest quartile (knee extension strength＜0.681kg/BW) was 3.07 (95%CI : 1.02-9.22) compared to highest quartile (＞0.980 kg/BW), while the multiple-adjusted OR was not significant (OR＝2.88, 95%CI : 0.94-8.79). However, the multiple-adjusted OR for fallers in the lowest quartile of the WS/SS ratio in knee extension strength (＜0.771) was 4.49 (95%CI : 1.40-14.37) compared to highest quartile (＞0.947).As for conclusion, without history of fall-related injuries, only a weak association between falls and physical fitness was observed in men. In women, the unbalanced knee extension strength was associated with falls. The WS/SS ratio in knee extension strength might be a key factor in predicting falls in women.