Free-burning experimental fires were conducted in a wind tunnel to explore the role of ignition type and thus fire spread mode on the resulting emissions profile from combustion of fine (< 6 mm in diameter) Eucalyptus litter fuels. Fires were burnt spreading with the wind (heading fire), perpendicular to the wind (flanking fire) and against the wind (backing fire). Greenhouse gas compounds (i.e. CO2, CH4 and N2O) and CO were quantified using off-axis integratedcavity-output spectroscopy. Emissions factors calculated using a carbon mass balance technique (along with statistical testing) showed that most of the carbon was emitted as CO2, with heading fires emitting 17 % more CO2 than flanking and 9.5 % more CO2 than backing fires, and about twice as much CO as flanking and backing fires. Heading fires had less than half as much carbon remaining in combustion residues. Statistically significant differences in CH4 and N2O emissions factors were not found with respect to fire spread mode. Emissions factors calculated per unit of dry fuel consumed showed that combustion phase (i.e. flaming or smouldering) had a statistically significant impact, with CO and N2O emissions increasing during smouldering combustion and CO2 emissions decreasing. Findings on the equivalence of different emissions factor reporting methods are discussed along with the impact of our results for emissions accounting and potential sampling biases associated with our work. The primary implication of this study is that prescribed fire practices could be modified to mitigate greenhouse gas emissions from forests by judicial use of ignition methods to induce flanking and backing fires over heading fires.