I don't know that I was surprised by the CLT itself. What surprised me about it was how nice it made everything else. In some sense there are whole books full of theorems which go from tedious and complex to 12-year-old simple in just 3 letters.

The Central Limit Theorem... that's a good one. Come to think of it, I felt enlightened by it even despite a teacher who wanted to be an actuary but had to teach AP Stat instead because she couldn't pass her actuarial exam, in addition to all the usual horrors bemoaned in the Lockhart essay that was posted the other day. I suppose that suggests that under more organic learning conditions, it would have been quite exciting.