OBJECTIVES: Recent research suggests that the value of health depends also on the time of permanence in a health state. This would imply a more complex relationship between quality and quantity of life than the standard linear relationship assumed in the QALY model. To model whether and how life duration affects the value assigned to health states. METHODS: A discrete choice analysis study was conducted comprising health-state scenarios added with a separate duration attribute. Health states were described with the EQ-5D (mobility, self-care, usual activities, pain/discomfort, anxiety/depression), having 3 levels of severity each. Duration  was introduced as a sixth domain with six levels (1, 5, 10, 15, 30, 50 years). Sixty choice sets were selected with a Bayesian approach (Stolk et al., Value in Health, 2010). A sample of 209 undergraduate students self-completed the computerized response tasks. Data were analyzed with a conditional logistic regression model. RESULTS: Main effects show negative preferences towards problems with health domains and positive preference toward longer duration. However, preferences are not linearly related with duration (e.g., the next 41st year is valued less than the next 11th year), but a logarithm function describes more accurately this relationship. Negative interactions are found between health-states and duration. Trends of values for health states with increasing duration are represented by diverging curves: when duration increases, the value assigned to good states increases while the value assigned to bad states decreases. However, due to the logarithm relationship with duration, the marginal (negative or positive) value assigned to any state decreases as duration increases. CONCLUSIONS: Duration of health states interacts with the value assigned to them and influences both direction and marginal values through the different durations. Our results show that refinement of the standard QALY framework can be amended.