Hormesis, the biphasic dose-response by which mild stressors trigger adaptive programs that leave cells better defended, has become the standard lens through which geroscience reads lifespan-extending interventions. However, the supporting evidence comes largely from young or middle-aged animals. This review argues that the hormetic window, i.e., the range of stimulus intensities an organism can convert into net benefit, is set not by chronological age but by integrated adaptive reserve, which is shaped jointly by aging biology, cumulative life-history exposures, and comorbidity burden. Because stress-response pathways across NRF2, HSF1, FOXO, AMPK, autophagy, and the telomere-associated DNA damage response are progressively attenuated, and because immune resilience governs the sensing, resolution, and repair phases of any adaptive response, the window both narrows and shifts toward lower doses. Doses clearly restorative in youth can therefore fall on the damaging side of the curve in old age, and two individuals of identical chronological age may respond very differently. We apply this framework to five paradigms: mitohormesis, caloric restriction, thermal stress, exercise, and genotoxic stress, and show that the stimulus generally retains its signal in old age while the downstream amplification machinery becomes rate-limiting. Interventions with explicitly age-dependent effects, including senolytics, NAD⁺ precursors, partial reprogramming, Bcl-xL overexpression, and stem-cell-derived extracellular vesicles, are examined as complementary probes of the same principle. The review closes with implications for geriatric trial design: comprehensive geriatric assessment combined with molecular biomarkers, longitudinal monitoring of a moving window, explicit safety thresholds, and sex-disaggregated analysis.