Snags (standing dead trees) are important components of forests that provide resources for numerous species of wildlife and contribute to decay dynamics and other ecological processes. Managers charged with managing populations of snags need information about standing rates of snags and factors influencing those rates, yet such data are limited for ponderosa pine (Pinus ponderosa) and especially mixed-conifer forests in the southwestern United States. We monitored standing rates of snags in 1-ha plots in Arizona mixed-conifer (n = 53 plots) and ponderosa pine (n = 60 plots) forests from 1997 through 2012. We used the Burnham live-dead, mark-resight model in Program MARK and multimodel inference to estimate standing rates during 5-year intervals while accounting for imperfect detection. Because snag standing rates may be influenced by plot characteristics, we used plots rather than snags as sampling units and conducted bootstrap analyses (500 iterations per model) to resample plots and estimate standing rates and associated parameters. We modeled standing rates in 3 discrete steps. First, we selected a parsimonious base model from a set of models including snag species, and then we evaluated models created by adding snag and plot covariates to the base model in steps 2 and 3, respectively. Snag standing rates differed among snag species and 5-year sampling intervals. Standing rates were positively related to snag diameter, negatively related to snag height, and were lower for snags with intact tops than for broken-topped snags. Standing rates also were positively related to topographic roughness, elevation, tree density, and an index of northness, and negatively related to slope and relative topographic exposure. Our results provide comparative data on standing rates of multiple species of snags based on a large and spatially extensive sample and rigorous analysis, and quantify the relative importance of several snag and plot characteristics on those rates. They indicate that modeling snag dynamics is complicated by both spatial and temporal variation in standing rates and identify areas where further work is needed to facilitate such modeling.