Rodent models of human diseases that accurately and reproducibly capture their pathology are key tools in furthering our understanding of the mechanisms behind these diseases and in the development of novel treatment approaches. However, pre-clinical studies in rodents are often criticised for the relative lack of replication and success upon translation to humans. Animal models of neurodegenerative diseases (and other CNS conditions) are very complex, often with multifactorial inputs into their development and progression. This complexity is a significant challenge. In addition to this, there are often concerns raised about the conduct, analysis and interpretation of the model results. This review focuses on Alzheimer's disease as a representative neurodegenerative disorder and will examine disease model end-points, in particular, behavioural phenotyping which, while appearing relatively straightforward, has the potential to be poorly conducted and the results misconstrued. This review uses a sample of the literature to illustrate the breadth of techniques used in behavioural assessment of Alzheimer's disease models, highlight the complexity, illustrate some procedural, interpretational and translational issues and provide recommendations to improve conduct of pre-clinical testing with the hope that this may lead to more consistency and translational success.