Chris Cillizza, in his blog for the Washington Post, lists 10 "battleground states", defined as the states most likely to switch away from the party whose presidential ticket they backed in 2004.
We tend to hear about the polls of the popular vote nationwide. But that, of course, doesn't really matter. As was not all that widely known before the 2000 election, but should be remembered by everyone who witnessed that contest, the only thing that matters is winning a plurality in enough states to give a candidate a majority of the electoral votes. Hence, the speculation about the individual states.
One problem I see with an analysis of which states will swtich sides, is that it might be based too much on the notion that the future will resemble the past. We're coming off two consecutive elections that were unusual in that they both had very close electoral vote margins, and only a handful of states switched sides between 2000 and 2004.
That's not to say that this year's result cannot be similar to those of 2000 and 2004. But we could very well return to the more-common pattern of recent decades, in which one candidate establishes a sizable lead in the popular vote, which translates into victories in states beyond his party's base, thereby producing a comfortable margin of victory in the electoral college.
After the conventions, I plan to write in more detail about the electoral college, and how the 2008 election is shaping up in "battleground" states.