Forest fires are large-scale disturbance events in many old world and neo tropic forests across the globe. Increased human pressures and altered climatic conditions have exacerbated the phenomenon, leading to rapid fire-return intervals in these forests. The study examined the spatial, temporal variability and the role of certain factors such as fuel loads and precipitation in influencing the pattern of fire occurrence across a landscape. The landscape is situated in southern India and spans 320 km², harbouring a number of vegetation types ranging from tropical dry thorn forests to tropical moist deciduous forests and associated with them is a strong rainfall gradient. Applying geographical information systems and remote sensing techniques, spatial data on fire occurrence for 14 years (1989 – 2002) was analyzed to generate a fire frequency map. An average of 30% of the sanctuary area has burnt over the 14 years of study. Areas in the tropical dry deciduous forests (64%) and tropical dry deciduous forest (Shorea type) (52.5%), have short mean fire-return intervals of 3-4 years and 1-2 years respectively. In contrast, areas in the tropical moist deciduous forest (degraded) (98%), tropical dry thorn (80%) and tropical moist deciduous forest (87%), have relatively longer mean fire-return intervals of 7 years. The data was modeled to capture the relationship between fire frequency, fuel loads, and precipitation across the landscape, providing a significant relationship (R²=0.63). We conclude that there is significant variability in the occurrence of fire across the different vegetation types. Precipitation and fuel loads are two important factors driving the behavior of fire in this region. The rapid fire-return intervals could have consequences for the ecology of plants and animals in the region.