Glycogen synthase kinase‐3 (GSK‐3) has been proposed as the main kinase able to aberrantly phosphorylate tau in Alzheimer's disease (AD) and related tauopathies, raising the possibility of designing novel therapeutic interventions for AD based on GSK‐3 inhibition. Lithium, a widely used drug for affective disorders, inhibits GSK‐3 at therapeutically relevant concentrations. Therefore, it was of great interest to test the possible protective effects of lithium in an AD animal model based on GSK‐3 overexpression. We had previously generated a double transgenic model, overexpressing GSK‐3β in a conditional manner, using the Tet‐off system and tau protein carrying a triple FTDP‐17 (frontotemporal dementia and parkinsonism linked to chromosome 17) mutation. This transgenic line shows tau hyperphosphorylation in hippocampal neurones accompanied by neurofibrillary tangles (NFTs). We used this transgenic model to address two issues: first, whether chronic lithium treatment is able to prevent the formation of aberrant tau aggregates that result from the overexpression of FTDP‐17 tau and GSK‐3β; second, whether lithium is able to change back already formed NFTs in aged animals. Our data suggest that progression of the tauopathy can be prevented by administration of lithium when the first signs of neuropathology appear. Furthermore, it is still possible to partially reverse tau pathology in advanced stages of the disease, although NFT‐like structures cannot be changed. The same results were obtained after shut‐down of GSK‐3β overexpression, supporting the possibility that GSK‐3 inhibition is not sufficient to reverse NFT‐like aggregates.