Ten years after the introduction of the first drug for the treatment of Alzheimer's disease, tacrine, it seems appropriate to reappraise the pharmacological processes of innovation in the field of research in dementia. The aim of this review is to pinpoint concrete improvements achieved in this field, in terms of experimental methods and clinical evaluation of the compounds, as well as the neurochemistry of the disease and cellular targets deserving of initial consideration. * The article first considers the use of animal models of Alzheimer's disease, which are classified according to two categories: animals with lesions of some neuronal pathways specifically implicated in clinical symptoms (i.e. lesions of the nucleus basalis of Meynert, the origin of cholinergic projections to the cortex underlying memory processes); and transgenic models, which are intended to reproduce some of the neuropathological hallmarks of Alzheimer's disease. Drugs can be tested in animals with such alterations for their effect on neuropathology, neurochemistry and behavioural disturbances. More recently, in silico models have been developed, which offer the possibility of simulating the pharmacodynamic effects of drugs in specific areas of the brain. These experiments are helpful in distinguishing purely symptomatic effects from disease-modifying effects, the latter being the ultimate goal of the modern pharmacology of dementia. * The second breakthrough considered in this article is the codification and standardisation of clinical methods for obtaining a more accurate and earlier diagnosis (the recent introduction of the concept of "Mild Cognitive Impairment", which includes patients who will later develop a true clinical dementia syndrome). In that respect, the determination of the biological markers of Alzheimer's disease (apolipoprotein E, amyloid substance, protein-tau, isoprostane) as well as progress in neuroimaging (functional positron emission tomography [fPET]-scan, single photon emission-computed tomography [SPECT], functional nuclear magnetic resonance [fNMR]) are discussed in terms of their potential as new tools in the early stages of drug development (surrogate markers). The methods used during the comparative clinical trials (phase III) have been elaborated and internationally standardised during the assessment of the different acetylcholinesterase inhibitors (AChE-I), with the knowledge that, since 1994, four of these have been officially approved: tacrine, donepezil, rivastigmine and galantamine; the same methods have been used for developing memantine, a recently-launched modulator of glutamatergic neurotransmission. The validated scales now take into consideration not only the cognitive dimensions of Alzheimer's disease but also the behavioural symptoms, with the introduction of the concept of BPSD (behavioural psychological symptoms of dementia). Some proposals to improve this clinical assessment of anti-dementia drugs are presented here. * The section of this article dealing with prospective issues considers the main pathways of interest in drug innovation and the elucidation of new targets for the future compounds. As well as their symptomatic effects on the different components of cognition, drugs should be neuroprotective and limit the lesions documented in Alzheimer's disease, with the aim of progressing far beyond the amyloid hypothesis (immunisation, beta-sheet breakers, secretase inhibitors). The field of excitotoxicity (which is mainly glutamate dependent) appears fruitful, because of the possibility of pharmacological intervention at the different steps in the excitotoxic process. All the new directions presented in this article support the concept of true disease-modifying agents. In conclusion, this prospective review should be considered as a guide in fostering drug innovation in Alzheimer's disease and related disorders and should help to decrease the gap existing between neuroscience and therapeutics.