OBJECTIVES: Health economic modelling has paid limited attention to incorporating the effects patients’ psychological characteristics can have on the effectiveness of a treatment.  The objective of this study was to test the feasibility of incorporating psychological prediction models of treatment response within an economic model of a diabetes structured education programme: Dose Adjustment For Normal Eating (DAFNE). METHODS: Data from the National Institute for Health Research DAFNE Research Programme were used to support all analyses.  Three regression models were used to investigate the relationships between patients’ baseline psychological characteristics and their 12-month HbA1c response to DAFNE.  The regression models were integrated with a patient-level simulation model of type 1 diabetes to evaluate the cost-effectiveness of two new policies (providing DAFNE only to predicted responders and offering a follow-up intervention to predicted non-responders) compared with current practice.  The model estimated costs and quality-adjusted life-years over a 50-year time horizon from a UK National Health Service perspective.  Deterministic sensitivity analyses were conducted. RESULTS: Psychological predictors of treatment response were successfully integrated with the health economic simulation model and allowed new treatment policies to be evaluated.  The results suggest that providing DAFNE only to predicted responders is dominated by current practice (incremental costs ranged from £297 to £616 and incremental QALYs from –0.112 to –0.209).  This result was insensitive to the psychological prediction model used and to the majority of sensitivity analysis assumptions tested.  The results suggest that providing a follow-up intervention to predicted non-responders dominates current practice.  This result was sensitive to model assumptions. CONCLUSIONS: By collecting data on psychological variables for a subgroup of patients before an intervention, we can construct predictive models of treatment response to behavioural interventions and incorporate these into health economic simulation models to investigate more complex treatment policies.  Further research using this methodology is indicated.