Personally I've not found this to be an issue (ok, maybe once or twice!), as I'm very used to camel casing from PHP and ActionScript.
But I agree it does introduce a margin for error.
For me it comes down to the convenience of being able to easily select the class/id name in my text editor, for copy/paste purposes. I weighed up the pros and cons of various naming conventions, and decided that for me the advantages of camel casing outweighed the negatives.
Another approach that would have met my requirements (double-clicking selects the entire class/id name), and eliminated any margin for error is if I used only lowercase, e.g.
In fact I used to name selectors in this manner, but found them confusing to read. After that I switched to hypenated selector names, but the double-click thing constantly tripped me up, so I eventually settled on camel case. Somewhere in between I might have used underscores too!
I guess there is no perfect approach. Unless my IDE has a hidden preference to ignore hypens when making a selection...
EDIT: I had been lead to believe that underscores were illegal in CSS selectors, as stated by Roger Johansson: "Class and id names can only consist of the characters [A-Za-z0-9] and hyphen (-)"). http://www.456bereastreet.com/lab/developing_with_web_standards/css/
However, reading the W3C section on CSS2 Syntax (http://www.w3.org/TR/CSS2/syndata.html#characters) it in fact states that underscores are acceptable.
Does anyone have thoughts on this?