Energy-related CO2 emissions embodied in international trade have been widely studied by researchers using the environmental input-output framework. In the literature two different approaches to deal with emissions embodied in a country's imports can be found. One of the approaches is based on the assumption of competitive imports while the other is based on the assumption of non-competitive imports. We show that the implications of the results obtained using different imports assumptions are not the same. The approach using the competitive imports assumption gives estimates larger than those obtained using the non-competitive import assumption. The differences between the two embodiment estimates come from the transition of embodied emissions in China's imports for intermediate consumption to those in China's exports. This explains why relatively high estimates of CO2 emissions embodied in China's exports are reported in several recent studies appearing in Energy Policy. (C) 2013 Elsevier Ltd. All rights reserved.