Trade and growth theories predict a mutual causation of innovation and exports. We test empirically whether innovation causes exports using a uniquely rich German micro dataset. Our instrumental-variable strategy identifies variation in innovative activity that is caused by specific impulses and obstacles reported by the firms, which can reasonably be viewed as exogenous to firms' export performance. We find that innovation attributable to this variation leads to an increase of roughly 7 percentage points in the export share of German manufacturing firms. The evidence is robust to several alternative specifications, similar for product and process innovations, and heterogeneous across sectors.