It looks like the EPA, GM and a few others are trying to figure out how to compare gas burning apples to plug in the wall oranges. Everyone knows what you are talking about when you quote MPG numbers but what about for a car that has gas tank or one that can be charged off the power grid?
It seems that the EPA is trying to come up with a standard for how to equate the two. Well just to throw my two cents in: equate them based on what really matters. No one really cares how much gas a car burns (except for concerns of range) what people really care about is either how much money it takes to go some distance or how much CO2 you dump out doing it. The first number would be a bit tricky as both fuel and electric prices very a bit by region so some kind of national average would need to be used. The second one is no better because CO2/kWh is a function of what is generating the reserve power on your grid? I happen to know that most hydro-power plants are run at either full power (or as close to it as something else lets them). The same almost surly goes for most of the renewable power sources as they tend to be use it or loose it systems. From that, all or most of the power production, will be from the dirtiest source around so it ends up being not at all trivial to compute.
Just out of curiosity, if a large chunk of the US fleet goes to plug in cars, how much power are we talking about? According to the linked article, 80% of cars go less than 40 mi/day and the only number cited for power consumption is 0.25 kWh/mi so if you assume the long tail puts the average at THE 80% mark, that gives you 10 kWh/day/car. If there is one electric car per 6 people in the US, that gives 50 GWhr/day or 1.5 TWhr/month of extra demand. For comparisons that is (based on mt reading of this report) about 0.1% of the total US electricity demand.