When do governments pursue unpopular reform, such as cutting benefits? And when do they engage in not-unpopular reform, such as activation? Current approaches in welfare state research cannot systematically explain the cross-government variation in the two types of reform. Based on insights from prospect theory, a psychological theory of choice under risk, this article complements existing theories by arguing that losses and gains matter crucially for welfare state reform. A fs/QCA analysis of labour market reforms pursued by 23 Danish, German, Dutch and British Cabinets between 1979 and 2005 corroborates this hypothesis. Specifically, it shows that an improving political position (a gain) is the necessary condition for not-unpopular reform while for unpopular reform it is a deteriorating socio-economic situation (a loss). This finding helps account for the puzzling cross-government variation in different types of welfare state reform © The Author(s), 2009.