The growing interest in analyzing the cost-effectiveness of interventions presents a challenge to anyone involved in dementia research. Although not yet often expressed as a formal requirement, drug authorities and budget holders also wish to have information on cost effectiveness. Clinical outcomes must therefore be combined with outcomes suitable for cost effectiveness analysis. However, issues about cost effectiveness are seldom included in empirical clinical trials, but are more often analyzed using modeling approaches. Clinical researchers and economists need to bridge these areas of potential conflict when long-term cost effectiveness is being considered. Randomized clinical trials (RCTs), observational studies, register data and economic models all have their advantages and drawbacks, and in making statements about cost effectiveness it is necessary to make a comprehensive judgment based on several methodological approaches. RCTs with a duration of at least 12 months should include assess-ments of resource utilization, and outcomes should offer a link to population-based cohort studies to discuss generalizability (e.g. a cognitive measure avail-able both in trials and cohort studies, such as the Mini-mental State Examination (MMSE), staging instruments, quality of life instruments) of the costs, while also serving as one source of modeling. Both dementia-specific and generic quality of life instruments are required. Models are necessary but must be transparent so that the assumptions on which the models are built can be critically analyzed.