Josh raises the question of whether a dragged-out Democratic primary will cripple the eventual nominee. (Though I'm not sure 1968 is a comforting analogy: Wasn't Humphrey-Nixon only close because George Wallace nabbed a bunch of Southern states from Nixon?) To go further, John Sides hauls out a study of presidential elections from 1936-1996, arguing that on-the-ground conditions in the country—especially the economy—have far more sway over electoral outcomes than intra-party fisticuffs. (Here's more evidence.) Still, that is a small sample, and it's easy to dream up reasons why this election might not be like past elections.
In a similar vein, David Greenberg points out in Slate that Democrats have frequently had brutal primaries, most of which make Obama v. Clinton seem downright cuddly in comparison, but that the nastiness doesn't have much effect come November:
Although the intraparty warfare sometimes got ugly in these races, and pundits warned of its harmful consequences, there's little evidence to suggest that it ever made a substantial difference in the fall election. In 1976 and 1992, the Democrats won. In 1972, 1980, and 1984, they surely would have lost anyway. In 1988, Dukakis met defeat because of his weak general-election campaign, not his springtime battles with Gore and Jackson.
Okay, but intuitively, Obama and McCain (say) seem much more evenly matched than McGovern and Nixon—or Mondale and Reagan. In a race this close, a small edge for one side could matter a great deal. A drawn-out dogfight means that the Dem nominee isn't going to be able to spend the next three months pouring $50 million into ads going after—and trying to define—McCain while he's struggling to raise cash, for starters. (Update: Err... a point Chris basically made below while I was sluggishly writing this post. I blame hunt-and-peck typing.)