If begun early in life, exercise effectively reduces the development of cognitive deficits in transgenic mouse models of Alzheimer's disease (AD). However, the effectiveness of exercise, once the cognitive impairments are established, is not as clear. In terms of translating research in animal models to treatments involving exercise in Alzheimer's disease patients, it is critical to evaluate exercise intervention at time points that address not only prevention, but also treatment of cognitive decline. We provided exercise wheels to Tg2576 (TG) (n=12) and C57BL6 (WT) (n=17) mice at 16-18 months of age for three weeks. At this age animals have significant cognitive impairment and neuropathology consistent with AD. Age matched sedentary TG (n=13) and WT (n=12) mice were also included, as well as groups provided access to an immobile wheel (TG n=9, WT n=12). After three weeks, animals were evaluated in a radial arm water maze. Significant impairments were observed in the sedentary TG mice compared to WT in reference/long-term and working/short-term memory, as well as in probe trials. Exercised TG mice demonstrated improvements in memory, which made them indistinguishable from WT mice on all tasks. In addition, animals provided with an immobile wheel exhibited improvement in some, but not all cognitive measures. Our findings demonstrate that exercise can improve cognitive performance in a mouse model of AD even if applied after the development of pathology.