Using annual data on 21 industrial countries from the period 1985 to 2009 and a large number of controls, this paper empirically analyzes the impact of technological change on unemployment. As proxy for technological change, it uses the ratio of triadic patent families to population. According to the regression results, an increase in technological change substantially increases unemployment over 3 years. There is no long-term effect, though. The results are robust to both endogeneity and numerous variations in specifications. They support theoretical contributions according to which faster technological progress may increase unemployment, at least during a transition period.