I'm pretty sure this information has been posted before, but let's look at the final RCP Senate averages and compare them to the last two cycles to see how we stack up (Presidential swing states and Senate seats deemed tossups). First off, in no race (26 total) over those two cycles has the republican outperformed his or her final RCP average, not once. This bodes well for democrats in NC and NH (and Orman in KS) who lead narrowly on the RCP average. In CO, IA, and AK the democrats trail by 2.5, 2.3, and 2.4 points respectively. This compares rather favorably to RCP's final average history as the median error in those 26 races favored the republican by 3.6 points. In fact, over 80% of these 26 races featured an error that favored the republican by 2.4 points or more. Looking at just the senate results over the last two cycles (13 races), the median republican lean was 4.1 points, with only one race favoring the republican by less than 2.5 points. This doesn't mean we are probably going to win CO, IA, AK, or KS, but it does mean that doing so wouldn't be unexpected or defy recent history in any way.
6:07 AM PT: To be clear, I don't think the democrats will retain the senate, but it wouldn't be that surprising if they did.. They may pull out one of those close races, most likely AK I think, but probably not CO or IA. I think the chances given by Dkos and Pollster (between 10 and 20 percent) seem about right.