Using a stacked differences-in-differences approach, we study the effects of Low Emission Zones (LEZs) in Germany. The implementation of stage 1 and 2 LEZs, which banned the most pollution-intensive vehicles from city centers, significantly reduced PM10 concentrations. The most restrictive third stage had no detectable, additional effect. Analyzing the mechanisms behind these improvements, we find weak evidence of a 2% traffic decline inside LEZs. Exploiting novel data, our main results document small but precisely estimated effects on the local fleet composition: LEZs induced the replacement of 50,000 older, emission-intensive diesel vehicles with newer, less polluting gasoline cars. Our estimates suggest that LEZs had lower social costs than previously estimated.