Behavioural disturbances are common in Alzheimer's disease and may be affected by medications being developed to enhance cognition or slow disease progression as well as by psychotropic agents developed specifically to affect behaviour. In many cases, Alzheimer's disease patients included in clinical trials are not selected for behavioural attributes and the patient population is heterogeneous at baseline with regard to these symptoms. Analyses of the behavioural data should include assessment of the effects of the agent on patients who were symptomatic at baseline as well as on the incidence of new behaviours in those without symptoms at the time of study initiation. Analyses may focus on symptomatic patients exhibiting specific degrees of improvement (e.g. 50% reduction in symptom severity). The analytic strategy chosen to characterize the behavioural changes occurring in clinical trials involving Alzheimer's disease patients will depend on the hypotheses being explored, the characteristics of the patients at baseline, the size of the population studied, the assessment methodology, and the outcomes of interest.