Are estimated here, based on various reports (HT - Andrew Sullivan). The people who put this interactive graphic together highlight the inconsistency of some of the reports (and there are a couple that clearly don't fit in with the rest of the estimates), but what struck me is that given how hard this is to estimate, how consistent these estimates actually were. Are basically looking at repeated claims of sixty civilian deaths, tops, with about half under Obama and half under Bush for the given date ranges. Ballparking is probably the best we can do on a covert program like this, and these multiple ballparks seem to give you roughly that answer (we can ignore the guy who said their have been zero civilian casualties as being simply wrong). We can make allowances for the fact that these estimates don't bring us completely up to date, and therefore it probably makes sense to tack some more civilian deaths onto that tally (although the accuracy of the strikes has almost certainly gone up over time).
By my count, this indicates that we've had 307 drone strikes. Let's suppose they're almost all legitimate targets (I know we disagree on al Awlaki, for example, but most of these guys aren't U.S. citizens and I assume even the harshest critics will grant that we're not just flipping a coin when we target people). That's 300 or so high value targets with - at most - 100 civilian casualties.
Beats the hell out of bombing Baghdad, if you ask me.
Can you imagine how different things would be if, after the invasion of Afghanistan, we had a fleet of these available to chase the guys that got away into the Pakistani mountains back in 2002, before we the blunder of invading Iraq? If we successfully prosecuted an extensive drone-based followup to the invasion of Afghanistan in 2002 Bush would have had much less evidence to stand on invading Iraq in 2003. He might have still gone forward, part of the reason why many people found that as palatable as they did was the fact that they still felt like the threat was at large.