OBJECTIVE: to identify subgroups within the population with reduced or delayed disability during healthy ageing. DESIGN: a longitudinal, community-based study. SETTING: Dubbo, New South Wales, Australia. PARTICIPANTS: 2805 men and women 60 years and older, first examined in 1988-89. OUTCOME MEASURES: activities of daily living assessed serially every 2 years over 8 years (scored in the range 0-6, least to most impaired); scores related to subsequent hospital admissions and to demographic, clinical and psychosocial characteristics at baseline. RESULTS: 1973 men and women provided complete follow-up data. Mean disability score at entry was low at 0.18 and increased to 0.69 by the final survey. Those having three or more hospital admissions (40% of the sample) had minimum disability (disability score approximately 0.3) around 5 years earlier than those with fewer admissions. Those with dementia or other mental illness had the most severe disability (mean disability scores of 3.15 and 2.13 respectively), but their numbers were very small. Those with a stroke or respiratory illness were more numerous and they had major physical disability (mean disability scores of 1.44 and 1.32 respectively). In a regression model, the statistically significant baseline predictors of disability at the final survey were age, body mass index, use of anti-hypertensive medication, history of stroke, depression score, peak expiratory flow and physical disability. CONCLUSIONS: the findings confirm reduced or delayed disability in older citizens requiring little or no hospitalization. Age, impaired peak expiratory flow and physical disability at study entry were most strongly predictive of disability, while stroke and respiratory illness were relatively common causes of severe disability.