Snags are used as roosting sites by many bats living in coniferous forests of western North America. Thus, providing sufficient numbers of snags both spatially and temporally in forested landscapes is critical to sustaining populations of these species. One aspect that remains poorly understood is length of time that roost snags persist on the landscape in a form suitable for use by bats. This information is critical for forest-planning efforts in ensuring long-term availability of snag resources on forested landscapes. We monitored condition of 339 snags used as roosting sites by long-legged myotis (Myotis volans) 1–5 years post-discovery from 2001 to 2006 across 6 watersheds in Washington, Oregon, and Idaho, USA. Persistence rates (i.e., probability a snag remains standing from year x to x + 1) of roost snags declined with year post-discovery in all study areas. Fir snags (Abies spp.) exhibited lower persistence rates than other conifer species. Data for the Washington area indicated only 4.3% of roost snags likely remain standing 10 years post-discovery, with half-lives of all snag species <3 roost-years. Model ranking of habitat models predicting fall year of roost snags revealed that snag condition models were the most parsimonious in all geographic locations. Roost snags larger in diameter, shorter in height, and with fewer branches on the bole were likely to persist for more years. These data indicate that snags used as roosts by long-legged myotis are suitable as roosting sites for only a few years before falling. We recommend management policies for coniferous forests in the Pacific Northwest, USA, that promote sufficient leave-trees in set-aside areas to provide for future suitable, large-diameter snags for bats in managed, forested landscapes. © 2012 The Wildlife Society.