OBJECTIVES: To determine whether self-reports of delayed care predict increased mortality and functional decline in community-dwelling elderly. DESIGN: Longitudinal cohort study. SETTING: Five counties in North Carolina. PARTICIPANTS: A total of 4,162 randomly sampled individuals aged 65 and older. MEASUREMENTS: The primary outcome was the proportional hazard ratio (HR) for death in cohorts stratified by self-reports of delayed or foregone care. A secondary outcome, functional decline, measured the cohorts' odds of developing increased dependency in activities of daily living (ADLs). Control variables included predisposing, enabling, and need factors. RESULTS: Of 3,964 eligible participants reporting, 61% never, 27% once in a while, and 12% quite often delayed care. Over 3 years, 13% of participants died, and 17% developed increased ADL dependency. Nevertheless, in unadjusted and adjusted models, neither 3-year mortality HRs nor the odds of functional decline differed between cohorts reporting varying degrees of delayed care. Survival probabilities remained higher for 15 years among those reporting delaying care often. CONCLUSION: The inability of self-reported delay to predict adverse outcomes in community-dwelling elderly suggests the need for better understanding and support of the care-seeking process and additional measures of timeliness of access.