I was really excited when Garmin came out with the 935, which uses wrist-based HR. Over the years, I have had a bunch of problems with chaffing with the chest strap, and I was very excited to be done with it, and not have to worry about putting a chest strap on.
All that said, I would still need to use the chest strap for biking. I use a Garmin 810 on the bike, and near as I can tell, there is no way for that to get HR from the 935. The good news is, since biking doesn't require nearly as much upper body movement as running, HR strap chaffing was never really a problem on the bike.
Some reading about wrist-based HR certainly raised some concerns for me. A lot of people said it was wildly inaccurate. When I got my 935, I ran looking at my HR zones, and they seemed to be about where they normally were, so I was pretty easily convinced that it was working well and accurate for me.
As I ran more though, there were a few times that I questioned the wrist-based HR. I thought about doing a comparison to the chest strap reading, but kept putting it off.
Finally one day, I decided that I had to compare, just for peace of mind. I put on my HR strap that is paired with my 810, and went for a run, hold my 910 in my hand, with HR displayed, and compared it with what I was seeing on the 935.
Much to my satisfaction, the numbers were pretty close to EXACT.
This is the graph comparing wrist-based versus chest strap:
The average and max numbers were exactly the same:
I was thrilled to death! I could finally put any doubts about wrist-based HR behind me right? No so fast!
Fast forward to winter. Since I'm a big skier, I do some non-traditional workouts up in ski country. One of these activities is called skinning. It involved sticking "skins" on the bottom of your skis, which allow you to walk uphill with your skis on. It's an awesome way to start the day on the mountain, and the trip up is a pretty good workout.
I had worn my HR strap on some earlier skins, and my HR was generally toward the upper zones when climbing, which made sense, since it is a lot of work. Just for fun, I turned on my 935 to track one of my skins up, and the HR data put me back on alert. During the climb up, I looked at the HR Zone readout, and it had me in high zone 1, which I absolutely don't believe.
The next data point that freaked me out a little was on one of my cold runs in New Hampshire. I saw a friend of mine put his watch on the outside of his jacket. This makes it easier to view and start/stop. I did this for one of my long runs, and was looking at my HR zone info during the run, which seemed like it was pretty accurate. Then I stopped to think... How could it possibly be measuring my wrist-based HR through 2 shirts and a jacket? The really strange thing is, it actually seemed to be pretty believable. As an added bonus, when I finished the run, my watch told me I had set a new record HR on my run!
In conclusion, I can't really say if the wrist-based HR is extremely accurate, or not. In general, it seems to work well for me, but I will always have a fair amount of skepticism, based on some of the strange things I have seen. My advice would always be to listen to your body, and see what your RPE tells you. This is the best way to do what's right for you.