American foulbrood is the most destructive brood disease of honeybees (Apis mellifera) globally. The absence of a repeatable, universal typing scheme for the causative bacterium Paenibacillus larvae has restricted our understanding of disease epidemiology. We have created the first multilocus sequence typing scheme (MLST) for P. larvae, which largely confirms the previous enterobacterial repetitive intergenic consensus (ERIC)–polymerase chain reaction (PCR)‐based typing scheme's divisions while providing added resolution and improved repeatability. We have used the new scheme to determine the distribution and biogeography of 294 samples of P. larvae from across six continents. We found that of the two most epidemiologically important ERIC types, ERIC I was more diverse than ERIC II. Analysis of the fixation index (FST) by distance suggested a significant relationship between genetic and geographic distance, suggesting that population structure exists in populations of P. larvae. Interestingly, this effect was only observed within the native range of the host and was absent in areas where international trade has moved honeybees and their disease. Correspondence analysis demonstrated similar sequence type (ST) distributions between native and non‐native countries and that ERIC I and II STs mainly have differing distributions. The new typing scheme facilitates epidemiological study of this costly disease of a key pollinator.