Snags (dead-standing trees) are biological legacies that remain after disturbances in forests. We enhanced the ecological underpinnings of snag management within the context of mixed-pine forest restoration in the northern Lake States by quantifying characteristics of live trees and snags within eighty-five 500-m 2 plots at Seney National Wildlife Refuge (NWR) in eastern Upper Michigan. Study plots represented reference conditions (i.e. no past harvesting, relatively unaltered fire regime) and altered (i.e. harvested, altered fire regime) conditions. We also compared three treatments for creating snags from live trees. Snags were found in 87% of the reference plots and 85% of the altered plots. The only snag variables that differed between plot types were mean snag basal area, which was greater in altered plots (Student's t-test, p = 0.04), and mean percent total basal area of snags (greater in reference plots, p = 0.06). The composition of snags differed only in the 10-and 25-cm diameter classes (Multi-Response Permutation Procedure, p < 0.10). The percentage of snags that developed into the most advanced decay class (DC) differed among treatments after 4 years (χ 2 = 16.49, p < 0.01), with 26% of girdled trees, 3% of prescribed fire trees, and zero topped trees reaching DC5. Logistic regression illustrated that the influence of predictor variables on DC development varied by species and treatment. The findings from this study, past studies, and ongoing projects at Seney NWR are directly applicable to innovative management of snags in mixed-pine forests.