Savanna fires contribute approximately 40-50% of total global annual biomass burning carbon emissions. Recent comparisons of emission factors from different savanna regions have highlighted the need for a regional approach to emission factor development, and better assessment of the drivers of the temporal and spatial variation in emission factors. This paper describes the results of open-path Fourier transform infrared (OP-FTIR) spectroscopic field measurements at 21 fires occurring in the tropical savannas of the Northern~Territory, Australia, within different vegetation assemblages and at different stages of the dry season. Spectra of infrared light passing through a long (22-70 m) open-path through ground-level smoke released from these fires were collected using an infrared lamp and a field-portable FTIR system. The IR spectra were used to retrieve the mole fractions of 14 different gases present within the smoke, and these measurements used to calculate the emission ratios and emission factors of the various gases emitted by the burning. Only a handful of previous emission factor measures are available specifically for the tropical savannas of Australia and here we present the first reported emission factors for methanol, acetic acid, and formic acid for this biome. Given the relatively large sample size, it was possible to study the potential causes of the within-biome variation of the derived emission factors. We find that the emission factors vary substantially between different savanna vegetation assemblages; with a majority of this variation being mirrored by variations in the modified combustion efficiency (MCE) of different vegetation classes. We conclude that a significant majority of the variation in the emission factor for trace gases can be explained by MCE, irrespective of vegetation class, as illustrated by variations in the calculated methane emission factor for different vegetation classes using data sub-set by different combustion efficiencies. Therefore, the selection of emission factors for emissions modelling purposes need not necessarily require detailed fuel type information, if data on MCE (e.g. from future spaceborne total column measurements) or a correlated variable were available. From measurements at 21 fires, we recommend the following emission factors for Australian tropical savanna fires (in grams of gas emitted per kilogram of dry fuel burned), which are our mean measured values: 1674 ± 56 g kg−1 of carbon dioxide; 87 ± 33 g kg−1 of carbon monoxide; 2.1 ± 1.2 g kg−1 of methane; 0.11 ± 0.04 g kg−1 of acetylene; 0.49 ± 0.22 g kg−1 of ethylene; 0.08 ± 0.05 g kg−1 of ethane; 1.57 ± 0.44 g kg−1 of formaldehyde; 1.06 ± 0.87 g kg−1 of methanol; 1.54 ± 0.64 g kg−1 of acetic acid; 0.16 ± 0.07 g kg−1 of formic acid; 0.53 ± 0.31 g kg−1 of hydrogen cyanide; and 0.70 ± 0.36 g kg−1 of ammonia. In a companion paper, similar techniques are used to characterise the emissions from Australian temperate forest fires.