The New York Times has made some really interesting calculations regarding the possible combinations of states that could lead to different particular outcomes. Specifically, Obama has some 431 (84%) different combinations that might result in a win, Romney has 76 (15%), and there are 5 (just under 1%) different routes that might lead to an electoral tie and, as a result, a vote in Congress.
This legitimates two pretty significant assumptions: (1) that the election is highly unlikely to result in an electoral tie though it does remain in the realm of possibility and (2) Romney has to tread a much thinner line than Obama - again, turn-out looks more and more and more like the critical factor.
ADDITION: I played with the protocols a little bit - NYT seems to indicate that if Obama can win Ohio and Florida the election is essentially a done deal. Worth noting for us election watchers.