The Kapalga fire experiment was designed to provide information for conservation management in northern Australia. It specifically deals with the regionally dominant savanna woodlands, so the results do not necessarily apply to other regional habitats, such as sandstone country, floodplains, and monsoon rainforest.

The experiment was established because fire is such an important part of the northern Australian environment, with half or more of the Top End burnt each year. Most of the fires are deliberately lit by people, including conservation managers. The dominant fire management practice is to burn extensively in the early dry season (May/June), when the country is still relatively moist, and fires tend to be low in intensity, patchy, and small in extent. This reduces the extent of higher intensity wildfires that inevitably occur later in the dry season.

Land managers and scientists all agree that fire needs to be actively managed in the Top End. The question is not 'Should the country be burnt?', but 'How (when and where) should it be burnt?'. The ecological effects of different fire regimes are inadequately understood, so there is uncertainty over precisely what fire regimes are best for nature conservation.

Savanna vegetation is commonly regarded as being highly resilient to fire, with subsequent regeneration occurring very quickly during the following wet season. However, the longer term effects of fire on vegetation have been poorly documented. We also know very little about the effects of fire on nutrient cycling, energy flow, and other important processes that are responsible for maintaining ecosystem health in the long term. Finally, we know little about the effects of fire on biodiversity. Some species of birds and small mammals appear to be declining in the region, and it has been suggested that changed fire regimes are causing these declines.

We can learn a lot about fire by making opportunistic observations on the effects of particular fires as they occur in the landscape. However, the most reliable way of determining the effects of different fire regimes is through an experiment, where the various ecological factors are controlled. In a controlled experiment, different responses can be confidently attributed to different treatments, and not to random variation or to differences in other factors that might not be apparent.

The main features of the Kapalga experiment are: (1) it was conducted at the landscape scale, with experimental units comprising 15-20 km2 catchments; (2) a range of fire regimes that currently occur in the region was tested; (3) the experiment was replicated, with each fire regime repeated at least three times; and (4) it involved researchers from a variety of universities and other organisations, as well as from CSIRO, covering a wide range of topics including fire behaviour, atmospheric chemistry, nutrient cycling, hydrology and stream dynamics, vegetation, arthropods, and all vertebrate groups.

Four fire regimes were tested at Kapalga: (1) Early - annual fires lit early each dry season (the most common contemporary management practice); (2) Progressive - annual fires lit progressively through the dry season as country dries out (this was considered to approximate traditional Aboriginal burning practices); (3) Late - annual fires lit late in the dry season (as commonly occurs in the region as uncontrolled wildfires); and (4) Unburnt - no fire. Experimental fires were lit for five years, from 1990 to 1994, with projects collecting baseline data for up to two years prior to experimental burning.