In this work, we address a forest management problem for timber production with fire concerns, employing a novel simulation-based optimization approach wherein forest management is iteratively guided by the feedback from fire spread simulations.The forest management problem involves selecting an alternative prescription for each stand, subject to various restrictions (e.g. bounds on ecosystems services), to maximize the net present value. For each stand, prescriptions involve projecting forest conditions and outcomes using species-specific growth and yield models, combined with different fuel treatment scenarios. In each iteration, the optimization problem is solved. Fire travel times between adjacent points in a grid representing the forest are calculated, based on fuel models associated with the selected prescriptions and other conditions as wind and slopes. Fire spread is simulated for all potential ignitions. Fire paths with a rate of spread greater than a given threshold are identified and constraints are added to the forestry problem to exclude their associated prescriptions to be jointly selected. This problem is re-optimized and the process is repeated until there are no such paths.We describe computational experiments in a Portuguese forest showing how trade-offs between the net present value and the maximum fire rate of spread can be obtained. When too restrictive conditions are imposed on fire, the approach suggests a set of stands to become fire breaks. We also conducted experiments to demonstrate how the impact of the forest surroundings, as well as bounds on ecosystem services, can be evaluated with respect to these trade-offs.