As a rule, they suck. Much, much harder than projecting hitters.
For pitchers with at least 100 innings last year, Chone came in at .424, way lower than the near .700 range for hitters. This represents the correlation coefficient between projected ERA and actual ERA. The other projection systems didn't do much better:
.445 Baseball Info Solutions (Bill James has nothing to do with this although its in his Handbook, he claims it can't be done)
Still waiting on the return of my 2006 Prospectus, so I can check PECOTA.
How about souting info? I tried putting together some projections based solely on a regression formula from scouting reports, with things like average fastball velocity, types of other pitches, quality of pitches, etc. Didn't do so hot:
.265. Of course, I'm no great scout, and a real scout attempting this with better scouting data might be able to do better.
The projections are better than using prior year ERA (.29) or prior year Fielding Independent Pitching (.37), though not much better than the latter.
I've heard some people claim that you shouldn't test a pitcher projection by looking at ERA anyway, you should look at component ERA or something. I don't buy that. I need ERA (or at least run average). Whether for fantasy baseball, where you need ERA to get results, or real baseball, where you need to keep real runs off the scoreboard, I don't give a damn if you can accurately predict component ERA. If you can't predict the real thing, then your system is useless.
Pitcher projections are a very inexact science. The only thing good I can say is as bad as they are, its better to have some flawed, not especially accurate idea of how a pitcher will do that to have no idea at all.