Electricity trade among power grids leads to difficulties in measuring greenhouse gas (GHG) emission factors of purchased electricity. Traditional methods assume either electricity purchased from a grid is entirely produced locally (Boundary I) or imported electricity is entirely produced by the exporting grid (Boundary II) (in fact a blend of electricity produced by many grids). Both methods ignore the fact that electricity can be indirectly traded between grids. Failing to capture such indirect electricity trade can underestimate or overestimate GHG emissions of purchased electricity in interconnected grid networks, potentially leading to incorrectly accounting for the effects of emission reduction policies involving purchased electricity. We propose a "Boundary III" framework to account for emissions both directly and indirectly caused by purchased electricity in interconnected gird networks. We use three case studies on a national grid network, an Eurasian Continent grid network, and North Europe grid network to demonstrate the proposed Boundary III emission factors. We found that the difference on GHG emissions of purchased electricity estimated using different emission factors can be considerably large. We suggest to standardize the choice of different emission factors based on how interconnected the local grid is with other grids. (C) 2015 Elsevier Ltd. All rights reserved.