Over the last decades social scientists have alleged that violence has decreased in Europe since late medieval times. They consider homicide rates a valid indicator for this claim. Thorough source criticism, however, raises serious doubts about the decline thesis having any substantial empirical foundation. Forms and contents of the sources are immensely heterogeneous and a closer look at the alleged richness of the data uncovers remarkable gaps. Furthermore, medieval and early modern population estimates are highly unreliable. Thus, we argue that historical research on violence should return to focus on specific historical constellations, accept the need for painstaking source criticism and pay careful attention to the contexts of violence.