Odd, isn't it, that something as significant to human health as HbA1c levels is so bloody hard to find consensus on? But Mendosa is right, there's surprisingly little information out there.
If non-diabetics aren't having the test done I suppose there must be very little data to work with? Even a sample of non-diabetics might not provide a truly representitive result, particularly in current circumstances where the majority of the population is perhaps eating more carbohydrate than at any previous time in our history? Normal today is almost certainly a good bit North of where normal would have been a few hundred years ago.
What would be most interesting would be a study of the HbA1c of indigenous tribespeople from around the world, people following diets that more closely resemble those of our ancestors. I'd guess they might be a bit lower than the 5's, but I've searched for the studies and drawn blanks so far.
Still, if the good 'ole ADA is going down to 6.5% for diabetics, it's probably safe to say the think 'normal' is a good bit lower than that.
fergus