The Cohen-Mansfield Agitation Inventory was a tool originally developed for use in research to measure agitation and subsequently used in clinical settings. It was the primary outcome measure for a randomised controlled trial which evaluated the effects and feasibility of a physiotherapist-led physical exercise intervention on agitation and physical performance of individuals living with dementia in nursing homes. The study produced weak results in regards to the Cohen-Mansfield Agitation Inventory due to small sample size and perceived issues with the use of the Cohen-Mansfield Agitation Inventory. Therefore, the focus of this paper is consideration of the identified issues by the research team: learnt effect and Hawthorn effect, misunderstanding of behaviours and inaccurate recall and observation. It is important that tools originally developed for research are still valid and reliable in the clinical setting.