In the 2015 refugee crisis, nearly one million refugees came to Germany, raising concern that crimes against natives would rise. Using novel county-level data, we study this question empirically in first-difference and 2SLS regressions. Our results do not support the view that Germans were victimized in greater numbers by refugees as measured by their rate of victimization in crimes with refugee suspects. Our findings are of great policy and public interest, and also of material relevance for the broader literature on immigration and crime which considers only crimes per capita or variants thereof, but never actual crimes by foreigners against natives. We show that this shortcoming can lead to biased inference.