This study examines underlying reasons for differences among land-based greenhouse gas flux estimates in Indonesia, where six national inventories reported average emissions of between 0.4 and 1.1 Gt CO2e yr-1 over the 2000 – 2012 period. The large range among estimates is only somewhat smaller than Indonesia's GHG mitigation commitment. To determine the reasons for these differences, we compared input data and estimation methods, including the definitions and assumptions used for setting accounting boundaries, including emitting activities, incorporating fluxes from various carbon pools, and handling legacy fluxes. We also tested the sensitivity of methodological differences by generating our own reference emissions estimate and iteratively modifying individual components of the inventory. We found that the largest changes stem from the inclusion of legacy GHG emissions due to peat drainage (which increased emissions by at least +94% compared to the reference), methane emissions due to peat fires (+35%), and GHG emissions from belowground biomass and necromass carbon pools (+61%), modifications to assumptions of the mass of fuel burnt in peat fire events (+88%), and accounting for regrowth following a deforestation event (-31%). These differences cumulatively explain more than half of the observed difference among inventory estimates. Understanding the various approaches to emissions estimation, and how these influence the magnitude of component GHG fluxes, is an important first step towards reconciling GHG inventories. The Indonesian government's success in achieving its mitigation goal will depend on its ability to measure progress and evaluate the effectiveness of abatement actions, for which reliable harmonized greenhouse gas inventories are an essential foundation.