A simple ecological model underlies contemporary fire policy in many West African countries. The model holds that the timing (or seasonality) of annual savanna fires is a principal determinant of vegetation cover. The model's origin can be traced to the ideas held by influential colonial scientists who viewed anthropogenic fire as a prime force of regional environmental degradation. The main evidence in support of the model derives from the results of a series of long-term burning experiments carried out during last century. The experimental results have been repeatedly mapped onto fire policy often taking the form of a three-tiered model in which fire exclusion is considered the ultimate management objective, late dry-season fire is discouraged and early dry-season fire is allowed but only under specific, often state-controlled circumstances. This paper provides a critique of contemporary fire policy in the region and the fire ecology model on which it is based. Through an analysis of burn scars for the 2002–3 fire season generated from ETM+ imagery, the study documents the spatiotemporal pattern of burning for an area in southern Mali. It argues that current policy, which is informed by an a-spatial model, cannot adequately account for the critical pattern of burning that is characteristic of the region. A reinterpretation of the burning experiments is presented in light of four factors: empirical data; recent developments in patch-mosaic theory; historical evidence on the effects of fire suppression; and data on indigenous burning strategies, all of which suggest a need to reconsider current fire policy.