This article revisits a property embedded in most dynamic macroeconomic models: the stationarity of hours worked. First, the author argues that, contrary to what is often believed, there are many reasons why hours could be nonstationary in those models, while preserving the property of balanced growth. Second, the author shows that the postwar evidence for most industrialized economies is clearly at odds with the assumption of stationary hours per capita. Third, he examines the implications of that evidence for the role of technology as a source of economic fluctuations in the G7 countries.