Despite an apparent structural integrity, the savanna landscapes of northern Australia are in flux. Important elements of the fauna have contracted in range and are less abundant than in the relatively recent past. Vegetation patterns are changing as populations of some important woody plants decline in some parts of the savannas while, in other places, different trees and shrubs are substantially increasing in density. These sorts of changes are occurring in lands under all tenures and subject to a variety of management goals and practices, including conservation reserves. Fire, large grazing animals and, more recently, invasive plants have all been implicated as drivers of adverse change. An important general, albeit inadequately tested, theory about apparently widespread faunal decline is that these influences have, jointly or separately, compromised landscape heterogeneity. It has been proposed that resource-rich patches that sustain savanna fauna through seasonal and longer-term peaks and troughs have been reduced in number, become more widely separated in space and time, or have been reduced in quality so that wildlife dependent on the rich patches struggle to use them effectively. The issues connected with these ideas were explored in a conference on fire and savanna management at Charles Darwin University in July 2003. Congruent with the theme of the symposium in which they were presented – Managing for Heterogeneity – the subset of papers presented here is diverse in origins, issues, perspectives and the spatial and temporal scales with which they deal. We consider that they make an important contribution to debate about conservation and development in northern Australia. Not because they answer the important questions, but because they illustrate the need for a shift in emphasis: from tentative exploration of pattern and weak inference about process to a harder-edged examination of the features of savanna habitats that influence their capacity to support viable wildlife populations.