Early 2016 Senate Ratings

[Vosem]

Alabama. Safe R.
Alaska. Safe R.
Arizona. Leans R. McCain is most likely out, which means a close race but probably also a narrow Republican victory -- pretty polarized, inelastic state.
Arkansas. Likely R. The state Democratic Party hasn't been exterminated yet like in other parts of the South, and Hillary at the top of the ticket could have powerful coattails. But on the whole, Boozman is pretty safe and has recently become very popular in the Republican Senate Caucus.
California. Safe D.
Colorado. Leans D. Swing state, Republican bench exists, but Republicans will be busy with defense and Bennet will have incumbency.
Connecticut. Safe D.
Florida. Leans R. Whether Rubio goes or not, Democrats will be competitive, but most likely will lose in a leans-R state to a good R candidate. With West this turns into Leans D, but I doubt he'll be the candidate if Rubio leaves.
Georgia. Safe R. Yeah, the state is trending, but Democrats have a very weak bench and Isakson is a strong enough incumbent to avoid this sort of trouble. Plus inelasticity.
Hawaii. Safe D.
Idaho. Safe R.
Illinois. Leans D. An uncontroversial Democratic candidate should be able to put this one away, but the Illinois Democrats are not known for not being controversial (don't like that sentence, too many negatives). Hedging my bets.
Indiana. Likely R. Coats is not a strong incumbent, but the Indiana Democrats don't have much of a bench.
Iowa. Likely R. Obviously Safe if Grassley runs, but I won't believe it till I see it. Tossup without Grassley.
Kansas. Safe R.
Kentucky. Leans R. Whether Paul runs for reelection or not, Democrats have a bench here but the state doesn't like electing Democrats federally. Depends on the top of the ticket, as well.
Louisiana. Safe R.
Maryland. Safe D. Mikulski could retire, but I don't think it'd matter.
Missouri. Leans R. Blunt vs. Nixon is Tossup, maybe even a shade of Tilt D; any other serious Democrat against Blunt is Likely R, verging on Safe. Averaged the two.
Nevada. Tossup. Reid vs. Sandoval is Tossup. Reid against anyone else is Likely D; Sandoval against anyone else is Likely R. Do the math.
New Hampshire. Tossup. Very elastic state. Ayotte seems to be doing OK now in my opinion, but this is a state where both parties seem to have enormous benches and she should receive a credible opponent.
New York. Safe D. I'd probably vote for Schumer myself.
North Carolina. Leans R. Tossup if it's Burr vs. Cooper, Likely R otherwise (though still competitive, Democrats do have a bench).
North Dakota. Safe R. Safe with Hoeven; Leans R if it's an open seat. Decided Hoeven is pretty likely to seek reelection, so Safe R.
Ohio. Leans R. Tossup if it's Portman vs. Cordray; Likely R if it's anyone else.
Oklahoma. Safe R.
Oregon. Safe D. I'd probably vote for Wyden myself as well.
Pennsylvania. Leans R. One of the few races where the matchup can be predicted with a good deal of confidence (Toomey vs. Sestak). And, yeah, in a good D year Sestak should win but otherwise Toomey seems to have impressed his constituents. Still, R Senator in an inelastic leans-D state, don't count out Sestak.
South Carolina. Safe R.
South Dakota. Safe R.
Utah. Safe R.
Vermont. Safe D.
Washington. Safe D. If the GOP didn't beat Murray in a midterm wave, 2016 can't work. She's on my retirement watchlist, but I doubt it'll matter. Republicans could decoy Democrats here, though.
Wisconsin. Tossup. Leans D if it's Johnson vs. Feingold (could be), Tossup if it's Johnson vs. Kind (doubtful), and Leans R if it's Johnson vs. anyone else (most likely). Averaged it out to Tossup.

Might make a map later, might not.

EDIT: And here it is:

To elaborate a bit further, the Republicans' problem isn't that of the Democrats of 2014, who are clearly about to lose several seats, but that they are so overstretched a couple of seats are bound to slip through the cracks and be Democratic victories. And the potential for gains, while there, seems to be even more miserly than the Democratic potential for gains in 2014 (though that might just be distance from the elections talking).

[Vosem]

who are clearly about to lose several seats

All right, slow down tiger. South Dakota is the only seat that's gone

West Virginia is as gone as South Dakota, Democrats are already behind in Montana and tied in Arkansas (red states where undecideds usually break against them); Begich barely leads Treadwell, far from 50, in an Alaska where Democrats consistently overpoll (and before you cite Miller he has not lead in a single poll), and in Louisiana, if Landrieu fails to break 50 she will have to deal with a runoff election where turnout will not be in her favor. It's quite a stretch to see Democrats winning three of these six.

and Kentucky and Georgia are on the table.

I'll address these separately. In Kentucky, Grimes is currently effectively tied with McConnell in the mid-40s, but Kentucky has been extremely averse to electing Democrats at the federal level recently and Grimes' campaign has been poorly managed (still time to fix that, but it's a bad early sign). In Georgia, polling has shown Nunn consistently in the low 40s, and like in Louisiana if she does not break 50 she will have to deal with an off-date election with low turnout. And Georgia is a more polarized state than, say, Indiana or Missouri -- it will be harder for Nunn to bounce off a gaffe or bad candidate to victory. In summary, Georgia is doubtful for Democrats but at least possible; while Kentucky won't fall without a strong regional or national wave for Democrats, neither of which is anywhere in sight.

The plausible range is from D+1 to R+6

To get D+1 Democrats need to sit down and get at least one more race seriously on the table. Natalie Tennant trailing in WV by double-digits doesn't count. The plausible range is somewhere from 0 (GA+KY/SD+WV) to around R+10ish (every Romney state and Michigan is clearly on the table, which makes 8, plus a maximum of 2 more from the 'a-hair-away-from-safe' quintet of CO/IA/MN/NH/OR; more than 2 of those is extremely doubtful).

, but realistically, it's gonna fall between R+1 and R+3

Realistically, Republicans are currently seriously shooting at 6 seats (AK, AR, LA, MT, SD, WV), while Democrats are shooting at 2 (GA, KY), so R+4 is probably the 'median' of possible futures. Democrats should be pretty satisfied with an R+3 result; that could correspond with small gains in the House.

Just because a democratic governor got elected doesn't mean this race is a potential pickup. Susana Martinez isn't going to block Tom Udall from sliding to victory in New Mexico is she? The right democrats can get elected in Missouri, but Blunt is the incumbent this time. The race is probably in his favor.

Martinez has never indicated any interest in the Senate, while Nixon actually ran for the Senate once already in 1998, doesn't seem to be planning to exit politics soon, won't be that old in 2016, and would have the Democratic nomination sewed up if he wants it, leading many to conclude that he is a possible 2016 Senate candidate.