External marking of American beavers (Castor canadensis) is essential to studies of population dynamics and behavior of individuals. Application of metal ear tags is a common method used to mark beavers but rates and causes of ear‐tag loss have been insufficiently documented. I live‐trapped and tagged 627 beavers in Voyageurs National Park, Minnesota, USA, from 2006 to 2012 with a single, uniquely numbered no. 3 monel ear tag in each ear. Beavers recaptured or recovered via live‐trapping (n = 178), legal harvest outside of the park (n = 55), or death (n = 4) up to 6 years after initial release were inspected for tag loss. An additional 31 beavers were secondarily marked with radiotransmitters and used to test the assumption that loss of one ear tag was independent of loss of the other. Six percent (15 out of 237) of beavers lost a single ear tag. Overall probability of beavers losing a single ear tag was low (3.3%) and did not differ by sex, age class, or method of recapture–recovery. Tag loss was highest within 6 months of initial tagging, but probability of tag loss was not related to time between tagging and recovery up to 6 years after tagging. None of the radioed beavers lost both ear tags, supporting the assumption of independence of ear‐tag loss among individual beavers. Ear‐tag loss rates are sufficiently low that studies of American beavers that rely on mark–recapture methods do not need to account for bias from ear‐tag loss. Published 2013. This article is a U.S. Government work and is in the public domain in the USA.