OBJECTIVE: To investigate the discrepancies between outcomes for competence (can do) and actual performance (do do) in activities of daily living (ADLs). DESIGN: Baseline measurements of a population‐based follow‐up study. SETTING: Leiden 85‐Plus Study, the Netherlands. PARTICIPANTS: Five hundred and ninety‐nine persons, age 85. The response rate was 86%. MEASUREMENTS: Face‐to‐face interviews. Measurements of competence and actual performance were based on the Groningen Activity Restriction Scale. Help received was assessed for several domains. Prevalence rates for disability were assessed according to the concepts of both competence and actual performance. Analysis was performed separately for basic activities of daily living (BADLs) and instrumental activities of daily living (IADLs). RESULTS: Seventy‐seven percent of the oldest old were competent to perform all the BADLs and performed them regularly. Fifteen percent were not competent to perform certain BADLs independently but performed them regularly with help from others. The prevalence of disability defined as inability in one or more BADLs was 22% for women and 10% for men. The prevalence of disability defined as inactivity in one or more BADLs was 16% for women and 17% for men. Only 5% of the oldest old were competent to perform all IADLs and performed them regularly. In spite of being competent, 70% did not perform certain IADLs regularly. The prevalence of disability defined as inability in one or more IADLs was 64% for women and 55% for men. The prevalence of disability defined as inactivity in one or more IADLs was 92% for women and 98% for men. CONCLUSION: The structural discrepancies between the outcomes of competence and actual performance have important consequences when estimating disability in old people. Promoting actual performance in IADLs may reduce disability.