Forest ecosystem management (EM) in North America has evolved from a theoretical concept to operational practice over the last two decades, but its implementation varies greatly among regions. This paper attempts to evaluate (1) if and how emulation of natural disturbances (END) is being used as a conceptual bases for implementing EM, and more particularly, what strategies are used to define the natural forest of reference, and (2) what temporal and spatial scale strategies are being considered for seven important retention elements (downed woody debris, snags, green trees, corridors, riparian buffers, large patches and old forest)? To conduct this evaluation, five guides from four geographically well-distributed regions in North America are compared. Although END is the central conceptual foundation underlying four of the five guides, a natural forest of reference is not always clearly identified and none of the guides consider future impacts due to global change. The major weakness common to all five guides is the lack of consideration of long-term forest dynamics, particularly the lack of clear strategies for retention elements at a temporal scale longer than a single rotation. Generally, the spatial scales chosen for retention elements are not well-justified ecologically and targets for each retention element are not identified at different spatial scales. We stress that strong efforts have been made to develop forest management that incorporates some elements of natural variability and which considers societal needs, but further improvements are required. We conclude by presenting some suggestions to improve the approach. For example, creating more realistic guidelines in integrating current and future forest dynamics with pre-settlement information and planning rotation lengths that are inspired by the dominant natural disturbance.