OBJECTIVES:  As the pharmaceutical industry increases its focus on treating chronic diseases, it is vitally important to obtain accurate estimates of future costs and benefits. To date, the standard discounted utility model (DU) has almost exclusively been used, even though many studies showed that time-inconsistent, hyperbolic discounting models better describe human behaviour. Such models are not currently being employed, mainly due to confusion and uncertainty about their underlying theory and applicability. The aim of this study is to highlight the differences between these two models when assessing health benefits over long time horizons. METHODS:  The difference in discounting was demonstrated using test data over a 15-year time horizon. The DU model used a constant discount rate over time, whereas the hyperbolic model contained two parameters, one for perception of time and one for the departure from the traditional model (Loewenstein and Prelec,1992). The parameters used to calculate the discount functions were those proposed by Cairns and van der Pol (2000). RESULTS: Not only did the QALY within each treatment group drop significantly but the total QALY gain under the hyperbolic model was 61% lower compared with the DU model. As a result, the DU model overestimated the added value of the intervention, whereas the hyperbolic model, keeping all other variables constant, led to a lower ICER. CONCLUSIONS:  As average life expectancy increases so does the financial burden of long-term interventions for chronic diseases. This necessitates a more refined assessment for reimbursement that more accurately reflects people’s behaviour. We have demonstrated that, depending on the values of the parameters, using a more valid discounting approach can dramatically affect the estimate of ICER. We conclude that researchers should not rely on the DU model for treatments that achieve long-term health effects as it fails to account for people’s changing behaviour as they get older.