This paper studies the effects of large-scale sport events with live spectators on COVID-19 infection trends at the local population level. Specifically, we compare the development of incidence rates in 41 German Nomenclature of Territorial Units for Statistics level 3 (NUTS-3) districts hosting a professional football match with at least 1,000 spectators vis-à-vis similar districts without hosting a match. Our empirical analysis builds on difference-in-difference and dynamic event study estimation for panel data. Synthetic control method is applied as a robustness check. While our findings generally do not point to significant treatment effects for the full sample of match locations, we find some noteworthy exceptions. Districts hosting first league matches with spectator attendance above the median (> 6,300 persons) and, particularly, matches without strict face mask requirements experienced a significant relative rise in incidence rates 14 days after the match. We also find that intra-district mobility increases on match days in treated districts, highlighting the significance of professional football matches as mobility-based infection transmission channel.