Correctly accounting for the emissions embodied in consumption and trade is essential to effective climate policy design. Robust methods are needed for both policy and research—for example, the assignment of border carbon adjustments (BCAs) and emissions reduction responsibilities rely on the consistency and accuracy of such estimates. This analysis investigates the potential magnitude and consequences of the bias present in estimates of emissions embodied in trade and consumption. To quantify the bias of embodied-emissions accounting, we compare the results from the disaggregated Global Trade Analysis Project (GTAP 8) data set which contains 57 sectors to results from different levels of aggregation of this dataset (3, 7, 16 and 26 sectors) using 5,000 randomly generated sectoral aggregation schemes as well as aggregations generated using several commonly-applied decisions rules. We find that some commonlyapplied decision rules for sectoral aggregation can produce a large bias. We further show that an aggregation scheme that clusters sectors according to their emissions and trade intensities can minimize bias in embodied emissions accounting at different levels of aggregation. This sectoral aggregation principle can be readily used in any input-output analysis and provide useful information for computable general equilibrium modeling exercises in which sector aggregation is necessary, although our findings suggest that, when possible, the most disaggregated data available should be used.