In a recent randomized duck blind study [JR Alpha, OneSpeedGo Institute, 2009], only 16.67% of waterfowl were found to be wearing protective headgear in potentially dangerous situations. This study did not consider feather color.
This site, cyclehelmets.org, presents data on both sides of the helmet discussion, leaning (it seems to me) more to the "helmets are hyped and probably not needed or useful, and may do more harm than good" school of thought, rather than the "wear them all the time to save you the one time you need it" school (me). [although I noticed that Hodgson is missing from the studies reviewed. Hmmm.] Bicycle Helmet Safety Institute appears to lean in the pro direction.
One line of reasoning that the anti- side seems to like to attack is the "it happened to me, so it must be true" anecdote. I agree that this is generally a rhetorical form that should set off all the BS sensors. On the other hand, when bicycle helmets do their job within their design envelopes (slow to mid-speed collisions), it seems to me they can turn what would have been the rare, yet personally significant brain injury, Natasha Richardson for example, into a non-event that is not reported in official injury statistics, and would only be available via personal anecdote. Most cyclists I know, after a fall at 20mph where they smack their helmet on the ground and get some bruises and road rash, would not report that to anyone. After an incident like that, it still seems very plausible to me that the helmet did make a critical difference, yet this outcome will remain unreported and not considered in all of the studies of accident statistics. (noted also that the "most cyclists I know" rhetorical form also triggers the BS sensors into red alert).
My friend lost control of his motorcycle on a decreasing-radius turn at over 80mph. He was wearing full gear. He was bruised and mashed up, and his bike was totaled, but he walked away with no serious injuries. This is an incident of the personal anecdote, the "protective gear saved me" pattern. I have no doubts that the conclusion is utterly valid and true. Does that mean the outcome would always be the same at that speed? Of course not. On the other hand, the anecdote is data to consider in your own personal safety decisions.
So at this point I look at the available data, combine it with personal experience (explicitly biased, to be sure), and try to reach a reasoned and convincing conclusion for myself: since I avoid high-speed auto traffic very intentionally, and maintain what I believe is a healthy perspective that cars are two ton steel death machines that cyclists should watch like a mouse would do in a cage full of hungry cats, and since 99% of my cycling occurs in the helmet design envelope, I'll keep wearing one. That doesn't mean I think everyone should, or that you need to. That's up to you.
If you often ride in the company of high-speed automobiles, or often ride faster than 25mph (or whatever the top limit of typical bike helmets is), or find yourself hitting the ground a lot because you like to huck monster cliffs, then a well-vented expanded polystyrene foam hat probably won't do much good for you, might be bad, and you may want to consider alternate forms of cranium protection better suited to your potential impacts. If you mostly ride slow in a pack of similarly attired and stylish commuters, and are so practiced and comfortable on your two-wheeled machine that the odds of a freak "I was just riding along at about 5mph, x happened, and I smacked my head into the pavement" incident seem vanishingly small to you, and the idea of wearing a helmet irritates you or does not mesh with your fashion sense, I also get it. Although you may want to reconsider something like balancing "vanishingly small" against possible permanent disability or death. So the picture is far from clear, the studies far from decisive, and the technology far from ideal. And vanity knows no bounds. Continued vigorous debate and flexibility is called for. We need more data, and better helmets. Biking is not nearly as risky as many people seem to believe, but it will still probably pay off to be safe out there. Get up. Go ride.