I know a lot of you are going to say "no duh" to this thread, but I post it only because I have heard the following statement a bunch of times; "no one can tell the difference between 256kbps (or put a bitrate there) and lossless."
I finally got off my arse and did some ABX testing to see if *I* could hear the difference. I used a MacBook Pro, and MacABX. I was using the DAC on a RSA predator to a set of AKG K701s as well as Sony MDR-7506s. I used a few tracks, but the one I spent the most time with was the Brazilian from the 2007 remaster of Genesis' Invisible Touch.
I did ABX testing with the following bitrates of MP3s: 128, 160, 192, 256, 320. The baseline was an Apple Lossless encoded version of the same track.
The results were clear, and surprised the heck out of me. I could hear the difference between a 128, 160, 192, and 256 kbps mp3 from lossless 100% of the time. This was using the methodology described on wikipedia about ABX testing. Using 10 samples per round, then letting my ears rest, etc.
Even more amazing (to me) was that I was able to differentiate the difference between a 320kpbs mp3 and the lossless file over 82% of the time. MacABX reported a less than a 1.9% chance that I was guessing.
This has really changed the way I view my encoding, and I will probably never use anything less than 320, but probably start using Apple Lossless or FLAC only!
p.s. My father who is 58 years old, and suffers from known hearing loss was able to discern 256 from lossless 100% of the time too. When he reach the 320 vs lossless tests, he scored less than 50% correct, and MacABX reported that he was guessing 100%.