The answer is yes, if recent trends hold up. Analysis of numbers from the last election vis-a-vis the polls at this time indicate the electoral math for Obama still depends on winning one state- Ohio.
Is the race closer than it appears?
Electoralvote.com has a database of state poll numbers for Kerry at this moment in time compared to how he did in the end. If that same trend happens for Obama, Obama loses Colorado, Florida, North Carolina and Indiana. New Hampshire is essentially tied and VA is only in Obama's category by 1.3%.
If Obama were to lose those (predictably in after a few recounts with the results not known until Thursday or Friday) it would all depend on Ohio. The good news is Ohio "only" tightens to 5.5%.
The average state went to Bush by 1.6% more than the polls suggested. Some important states were more significantly off: New Hampshire & NV both 7+ underestimated Bush support. Combined with the primary polls, and New Hampshire certainly has to be considered a toss-up.
There are many reasons suggested for this which indicate Obama has nothing to worry about- the "October surprise" of the Bin Laden tapes, Swift Boats, large base turnout. We are also talking about candidate Kerry who certainly isn’t candidate Obama and never had the cash to host a half-hour variety show. Also, what about the new voter turnout, GOTV efforts and cell phone bias? These could certainly counter-balance or change the numbers from last time.
However, if the black vote is "locked in" and accounted for as some have suggested, if there is some Bradley effect, if PA is within McCain's reach and tighter than it looks as Governor Rendell has suggested, and if there's a reason McCain & Obama have been spending time in Iowa things could be more interesting than a lot of people are expecting.
More data would prove to be helpful on making a more accurate prediction, but there's plenty of reason for the Republicans to not give up hope and for Democrats to push as hard as they can to close the deal.