Due to Alzheimer's disease (AD) great aggressiveness, many worldwide health associations began to globalize research efforts in order to find a suitable treatment and to clarify once and for all its controversial aetiology. Moreover, the animal modelling research is one of the best tools to evaluate molecular mechanisms and to correlate them with clinical features and behaviours. However, in order to provide valuable scientific data correlated to low error sources, a rigorous algorithm of selecting the proper animal model for testing is required. An ideal animal model for AD research has probably not yet been developed, but by a careful selection of the existent models or even by developing new models suitable to research conditions, consistent progress in this area of research can be achieved. This paper aims to show and centralize some of the valuable information gathered along the past years of failure and success in Alzheimer's disease animal modelling, in order to provide a theoretical ground for new and innovative aspects in this rather new area of research.