Objectives: The main objective of this study is to conduct a systematic review to identify and discuss methodological issues surrounding decision modelling for economic evaluation of non-pharmacological interventions (NPIs) in dementia. Methods: A systematic search was conducted for publications using decision modelling to investigate the cost-effectiveness of NPIs for individuals with dementia. Search was limited to studies in English. Studies were excluded if they evaluated interventions aimed only at caregivers of patients with dementia, or if they only included economic evaluation alongside an RCT without additional modelling. Results: Two primary, five secondary and three tertiary prevention intervention studies were identified and reviewed. Five studies utilised Markov models, with others using discrete event, regression-based simulation, and decision tree approaches. A number of challenging methodological issues were identified, including the use of MMSE-score as the main outcome measure, limited number of strategies compared, restricted time horizons, and limited or dated data on dementia onset, progression and mortality. Only one of the three tertiary prevention studies explicitly considered the effectiveness of pharmacological therapies alongside their intervention. Conclusions: Economic evaluations of NPIs in dementia should utilise purposefully-developed decision models, and avoid models for evaluation of pharmaceuticals. Broader outcome measures could be a way to capture the wide impact of NPIs for dementia in future decision models. It is also important to account for the effects of pharmacological therapies alongside the NPIs in economic evaluations. Access to more localised and up-to-date data on dementia onset, progression and mortality is a priority for accurate prediction.