Fire managers currently use simple elliptical models to predict the perimeter of a fire when the fire starts from a single point. However, when examined closely wildland fire perimeters are highly irregular. We tested the hypothesis that a fire is actually fractal in nature and thus the true length of a fire perimeter depends on the amount of fine edge detail included in the measurement. The amount of perimeter detail incorporated is dependent on the length of the base unit of measurement; the longer the unit, the less the perimeter detail, and the shorter the perimeter. Different forest fire suppression techniques have inherent scale characteristics that are associated with different base unit measurements of perimeter, therefore, the length of the fire perimeter is dependent on the suppression technique used. Analysis of 14 forest-fire perimeters yielded a consistent fractal dimension of 1.15, and a relationship was found to adjust predicted fire perimeter with this fractal dimension for a specific unit measurement length. The fractal length of fire fronts between two identifiable points can also be calculated given the base unit measurement of the suppression technique. This information should improve fire managers' ability to optimally dispatch suppression resources to forest fires. © National Research Council of Canada, NRC Research Press. Abstract reproduced by permission.