According to the β-amyloid (Aβ) hypothesis, compounds that inhibit or modulate γ secretase, the pivotal enzyme that generates Aβ, are potential therapeutics for Alzheimer's disease (AD). Studies in both transgenic and non-transgenic animal models of AD have indicated that γ-secretase inhibitors, administered by the oral route, are able to lower brain Aβ concentrations. However, scanty data are available on the effects of these compounds on brain Aβ deposition after chronic administration. Behavioral studies are also scarce with only one study indicating positive cognitive effects of a peptidomimetic compound acutely administered (DAPT). γ-Secretase inhibitors may cause abnormalities in the gastrointestinal tract, thymus, spleen and skin in experimental animals and in man. These toxic effects are likely due to inhibition of Notch cleavage, a transmembrane receptor involved in regulating cell-fate decisions. Some non-steroidal anti-inflammatory drugs (NSAIDs) and other small organic molecules have been found to modulate γ secretase shifting its cleavage activity from longer to shorter β-amyloid species without affecting Notch cleavage. Long-term histopathological and behavioral animal studies are available with these NSAIDs (mainly ibuprofen) but it is unclear if the observed in vivo effects on Aβ brain pathology and learning depend on their activity on γ-secretase or on other biological targets. The most studied γ-secretase inhibitor, semagacestat (LY-450139), was shown to dose-dependently decrease the generation of Aβ in the cerebrospinal fluid of healthy humans. Unfortunately, two large Phase 3 clinical trials of semagacestat in mild-to-moderate AD patients were prematurely interrupted because of the observation of detrimental effects on cognition and functionality in patients receiving the drug compared to those receiving placebo. These detrimental effects were mainly ascribed to the inhibition of Notch processing and to the accumulation of the neurotoxic precursor of Aβ (the carboxy-terminal fragment of APP, or CTFβ) resulting from the block of the γ-secretase cleavage activity on APP. Two large Phase 3 studies in mild AD patients with tarenflurbil (R flurbiprofen), a putative γ-secretase modulator, were also completely negative. The failure of tarenflurbil was ascribed to low potency and brain penetration. New Notch-sparing γ-secretase inhibitors and more potent, more brain penetrant γ-secretase modulators are being developed with the hope of overcoming the previous setbacks.