OBJECTIVES: The United Kingdom Prospective Diabetes Study (UKPDS) reported a persistence in risk reduction of diabetes-related events associated with improved glycaemic control observed between intensive and conventional therapy groups beyond the intervention period. This has important implications for projecting short-term clinical trial results over long-term time horizons.  The aim of this study was to reproduce the UKPDS legacy effect and assess the impact on long-term cost-effectiveness. METHODS: The Cardiff Type 2 Diabetes Model was initiated with cohort profiles consistent with reported intensive versus conventional control groups within UKPDS; initial HbA1c treatment effects were applied and modelled over time assuming two scenarios: a loss of antihyperglycaemia benefit at year 10 or maintenance of clinical benefit (the legacy effect).  Under both scenarios, risk reductions and cost-effectiveness of sulphonylurea (SU) versus insulin were assessed over a 40-year time horizon using UK 2010 costs. Both costs and health benefits were discounted at 3.5%. RESULTS: The risk ratio (RR) of any diabetes-related end point predicted by the model was consistent with that reported by UKPDS when incorporating the legacy effect (RR of 0.90 versus 0.91 in the model and UKPDS, respectively).  Ignoring the legacy effect resulted in a RR of 0.99 at year 30 and a cost per quality-adjusted life-year (QALY) of £162,400, compared with £22,565 when including the legacy effect. CONCLUSIONS: The legacy effect of intensive glucose-lowering strategies has important implications when assessing the cost-effectiveness of new therapies.  Failure to include such a legacy effect, as seen in UKPDS, may result in new therapies for managing glycaemic control appearing less cost-effective than they actually are.