One strong possibility is delegate deadlock until the convention.  I wrote up this scenario on Daily Kos.  For the Democratic party there are 4,000 delegates at the convention.  

~800 are not elected by voters directly, but are delegates by virtue of their position as elected officials or party leaders.  ~3200 are elected in primary elections.  Of this 66% will have been chosen by Feburary 5th.

In order to get delegates a candidate must get more than 15% of the vote.  There's a complex procedure that involves at large delegates based on the state vote, and delegates awarded at the level of Congressional districts. The bottom line being that in order to win without needing superdelegates on the first ballot, a candidate has to have ~63% of the elected delegates. So the scenario is like this:

As it stands now in the elected delegate count, Obama has 25 (37.3%), Clinton 24 (35.8%), and Edwards 18 (26.9%). Only 2% of the elected delegates have been chosen at this time, so the situation is fluid, however the February 5th primaries will provide 1,698 elected delegates (52.3%), and  by that date a cumulative 2,148 elected delegates (66.1%) will have been awarded. Assuming that in the long run, Obama carries 38% of the elected delegates, Clinton 38%, and Edwards 24%, this would give the following delegate count:  Obama 1,234 Clinton 1,234 Edwards 780.  Again to illustrate this graphically, see the following. In order to win on a first ballot with 2,020 delegates, either Clinton or Obama would have to win more than 99% of the superdelegates.

This seemed highly unlikely a few days ago, but now we have these polls coming out of Nevada suggesting that Edwards has experienced a meteoric rise there, the question is why.  One plausible explanation is that Edwards is picking up disaffected white males, and more importantly hispanics.  In the US, there are important issues of race that divide along Hispanic/Black lines where there is a great deal of mutual animosity between the two groups.  If this is the source of Edwards rise in the polls, then we could have a real suprise in store.

Because it most likely underestimates Edwards strength, because in the absence of crosstabs based on language of the interview, I'm inclined to beleive that Spanish speaking Hispanics are underrepresented.  Which means that Edwards could pull off a win in Nevada.  And that would give him an important boost going into his native South Carolina, this being the only state he actually won in 2004.

Appreciate the irony of this.  For all the talk of race and gender in this election cycle, the two states brought in early to increase racial diversity in the starting states could produce wins for a white male.  Which will surely send the pundits into meltdown. And if it is Hispanics disaffected by the racial focus in this election cycle, then Edwards just might walk away from Feb 5th with close 2nd placings in California and the Southwest.

And that could very easily yield a situation in which there is delegate deadlock, and there will be no Democratic nominee until August.......

And I'll give my consent to any government that does not deny a man a living wage-Billy Bragg

What is striking about these figures is just how open both these races still are.   It is almost as if the voters are saying they are not ready yet to make up their minds and are trying to keep their options open for as long as possible.

One of the more annoying habits of elections commentary is ascribing a collective will to the electorate. It would be possible in large caucus, where a consensus might be reached that "well, we want to keep the race open, let's throw this percentage of our votes to that one" or something similar.

But with a secret ballot, no public debate between voters, and indeed that voters only vote once, an electorate doesn't 'decide to give a narrow victory' or whatever.

Un roi sans divertissement est un homme plein de misères

Previously contested conventions were the norm. At least that made the conventions more interesting. I am not sure that taking a few ballots to come up with a nominee is going to damage a party.

The only thing damaging to a party is if the convention fight is so long and bitter that it proves that the party is badly split. The 1924 Democratic convention, when William G. McAdoo (backed by rural, southern and western delegates, who were dry on prohibition) and Alfred E. Smith (urban, Catholic and wet on prohibition) battled it out for more than 100 ballots before the weary convention settled on a compromise candidate. Of course the severity of the deadlock was made worse because a two-thirds vote was then required to nominate.