Chris Cillizza of the Washington Post has a cogent analysis of Mitt Romney's troubles with the electoral college versus Obama this fall:

A detailed analysis of Romney's various paths to the 270 electoral votes he would need to claim the presidency suggests he has a ceiling of somewhere right around 290 electoral votes.

While Romney's team would absolutely take a 290-electoral-vote victory, that means he has only 20 electoral votes to play with - a paper-thin margin for error.

Romney's relatively low electoral-vote ceiling isn't unique to him. No Republican presidential nominee has received more than 300 electoral votes in more than two decades. (Vice President George H.W. Bush won 426 electoral votes in his 1988 victory over Massachusetts Gov. Michael Dukakis.)

[...]

In 2000, Bush won 271 electoral votes - one more than he needed to claim the presidency. In eking out that victory, Bush not only carried the South and Plains states with a near sweep but also claimed wins in swing states such as Nevada, Colorado, Missouri and the major electoral-vote prizes of Ohio and Florida.

If Romney was able to duplicate Bush's 2000 map, he would take 285 electoral votes - thanks to redistricting gains over the past decade.

But to do so, Romney would need not only to win the five swing states mentioned above - with the exception of Missouri, all of them are considered tossups (at worst) for the president at the moment - but also hang on to states such as North Carolina and Virginia where Bush cruised 12 years ago. (Obama carried both states in 2008.)

North Carolina, Virginia, Nevada, New Mexico, Colorado, and Missouri; these are the states that will be pivotal for Romney to win if he expects victory in November.

There are some other, more unlikely scenarios including a Romney win in Pennsylvania, or even New Jersey if he were to choose Chris Christie as vice president. If either one of those states were to fall to Romney, it would allow him to lose Virginia and New Mexico and still come out on top.

North Carolina is close now but probably won't be by election day. And Mormons in Nevada will probably make the difference for Mitt in that state.

But Romney finds himself trailing in Virginia at the moment as the state's demographics continue to shift unfavorably for him. Northern Virginia's share of the state vote continues to grow at the expense of more Republican districts further south.

Missouri has a hot senate race which may favor Romney on election day, but New Mexico and Colorado, with a large influx of Hispanics in both states, grow bluer by the year.

A realistic victory scenario for Romney probably includes an unexpected win in a state like Michigan or Wisconsin. The upper midwest is still hurting economically and if Romney can sell his economic program, he has a chance to eke out a victory in one of those two states.

As Cillizza points out, though, Romney also has a solid floor of about 190 electoral votes. That will help as resources are allocated in the last two weeks of the campaign not to defend home turf as McCain was forced to do, but to concentrate entirely on swing states and targets of opportunity.

It's a narrow window for Romney to be sure, but far from impossible.