On electoral college maps Obama maintains a slight lead, but his lead has eroded enormously in the last couple of weeks, and he doesn't have the 270 electoral college votes needed to win. Cook Political Report calls it a toss-up. Rasmussen shows that Obama's previous 210-165 advantage has dwindled to 193-183. (The site Electoral-Vote.com puts Obama over the top at 278 electoral college votes, but they count people who are "barely likely" to vote for Obama, and they appear to be a partisan site.) Pollster.com shows Obama with a larger lead, but it doesn't say which dates it is basing its numbers on.
In terms of national polls there isn't a clear trend; since July Obama has stayed slightly ahead of McCain. In this graph from Real Clear Politics, if you put your cursor on July 1 and draw an imaginary line to today, Obama is flat and McCain is trending up. But if you run your cursor along the timeline to see the difference in polling, it's a pretty steady, small advantage for Obama.
Obama is doing a lot better on the election prediction trading sites. On InTrade and Rasmussen markets he's currently selling at 60, compared to 40 for McCain.
The favorability ratings of the two candidates are virtually identical, and the support they have from their base is also nearly the same (link). The Democrats had a lot more voters in the primaries than the Republicans did, which means they have more voters registered and may do better in getting out the vote; but in recent elections the Republicans have beat the Democrats on that score, and I haven't heard any details on how that's going.
The Moderate Voice points out that the Republicans are yet to see their own convention bounce, and also that the surprise element of the Palin pick means that she could be a positive or a negative as more is learned about her. (HuffPo is throwing the kitchen sink at her, a tactic that might backfire as a number of the attacks are turning out to be scurrilous.) According to Gallup, we can expect McCain to get about a 5% bounce after his convention.
By tomorrow all the numbers will have changed, and every day is a better predictor than the last. As David Moore points out, pollsters ask people how they would vote if the election were held today, which provides shaky results until you get pretty close to the election. The only thing that seems safe to say at this point is that it will be close.
Another thing... While I was researching this post I stumbled on this article, which questions the accuracy of a CNN poll saying that an increasing number of Hillary supporters are switching to McCain. Since my own support for Obama is increasingly wobbly it feels in my gut like CNN was correct, but then I'm just a sample size of one.