There was one where the guy behind it went to massive lengths so people couldn't easily distinguish the example files by other means than audio quality. Verdict was that people with more expensive equipment even preferred the sound of the MP3s (320kbps CBR). I think it was this one (Links to Parts 2 and 3 at the bottom.).

Somewhere else I've read that - for most humans - 256 kbps MP3s encoded with VBR-ABR using a high-end encoder are basically indistinguishable from the lossless original. Even at 192 kbps it's still more hit&miss than it should be. But I don't remember where I've read that.