There's an interesting research done on this. I can't remember the name of the effect; perhaps someone mentioned it before me or could remind me? The basic theory is that the more protection one has, the more danger they feel confident with and procede with, therefore negating the effects of the aforementioned protection.
The simple act of not wearing a helmet cautions the drivers that this cyclist is quite possibly an idiot and dangerous. (I am one of such peoples) This advises drivers to stay farther away from said cyclists and allow more space when passing. This research observed that cyclists who wore helmets had cars that passed by closer.
I wear my helmet when I'm going for a real fast ride. I can go 50km + on flat roads, and 60km + when going downhill. If I intend to go at these speeds, I put on a helmet. Yet if I go at a speed of 10km ~ for a trip to the store, why would I need a damned helmet?
I also put on my helmet when I know I'm going to be working around kids. (I volunteered at a Day Camp for three weeks and wore a helmet everyday there. Nothing like teaching kids to be safe)