Background Reserves' mechanisms explain inconsistencies between accumulation of neuropathological damage and clinical manifestations. Leisure activities are believed to promote reserves. Objective This study evaluates whether cognitive, social, and physical leisure activities performed over life-span predict current cognitive functioning in normal aging and Alzheimer's disease (AD) continuum. Methods 35 AD, 24 amnestic-Mild Cognitive Impairment (a-MCI) patients, 21 individuals with subjective cognitive complaint (SCD), and 25 controls underwent a questionnaire developed to quantify leisure activities in different life periods, the Addenbrooke's Cognitive Examination-Revised (ACE-R), and T1-weighted 3T-MRI scans for brain volumetrics and cortical thickness quantification. Partial/total leisure activities' scores and demographic and brain variables were entered as predictors, while ACE-R scores as dependent variables in linear regression analyses. Results Current level of cognition was predicted by (i) social and physical activities performed in middle age and current cognitive activity in AD; (ii) cognitive and social activities performed in middle age, current age and cortical thickness in a-MCI; (iii) recreational activities the set of lifetime, current age, and brain features in SCD; (iv) education and the set of lifetime leisure activities over lifespan in controls. Conclusions This study shows a funnel effect due to gradual reduction of stimulatory activities in the transition from healthy aging to AD. Reserve indices taking into account different types of stimulatory activities allow to capture even smallest residual effects of reserves accumulated over lifespan, until their complete depletion at advanced AD stages. These results may help target tailored interventions during normal and pathological aging.