Contrary to common belief, polls actually offer a reasonable window into outcomes and generally are the best way to understand election trends. For example a national CNN poll in October of 2000 showed Bush with a slight lead over Gore and he wound up winning. This is not a great example because Gore did narrowly win the popular vote, but the point is that the CNN poll showed Gore and Bush about equal in October and they were about equal in the final vote.
Almost all the latest national polls show Obama with a lead of about 4-5%, though I don't think the debate's effect on things - if any - is reflected in any of them. I'm in the camp that believes this is now Obama's election to lose. Obama is ahead by enough and the debate seemed even enough that it will be difficult for McCain to gain more than few points barring the kind of events that now appear unlikely. Although the Palin / Biden VP debate is very anxiously anticipated I doubt that Palin is going to score any points for McCain, and the final two debates with Obama are likely to go the same way as the first - where the candidates appeal mostly to their own constituencies with McCain appearing to lose some points for his "angry" attitude, which I think helps explain why most non committed voters thought Obama won last night.