In the standard gamble and time trade-off methods of health state utility assessment, a specified health state and an alternative are compared. This alternative can be framed in terms of a loss or a gain in reference to the first health state. In this paper, we test whether this framing affects the estimated health state utilities. The experiment was carried out on a group of pharmacy students, randomly divided between the loss or gain version (n = 182). The null hypothesis of no difference between the loss and gain versions is rejected for the standard gamble method, but not for the time trade-off method.