We used data on 2356 trees from 43 prescribed fires in Idaho, Montana, Oregon and Washington states to model postfire tree mortality. Data were combined for seven species of conifers to develop binary logistic regression models for predicting the probability of mortality. Probability of mortality increased with percentage of the crown killed, and decreased as bark thickness increased. Models are presented with and without species as a categorical variable. The models predicted well for trees burned in both slash fires and fires in natural fuels. The models are applicable for assessing fire-caused mortality both of individual trees and in mixed conifer stands of the Pacific Northwest. © National Research Council of Canada, NRC Research Press. Abstract reproduced by permission.