Interesting. Cognitive Daily: Few listeners can distinguish between "average" and "best" MP3 samples Summary: Their results: 64 kbps worse than 128 kbps, 128 kbps usually not distinguished from 256 kbps. Audiophiles slightly better on testing than non-audiophiles. The test metholodolgy is much-maligned in some quarters, because they did not use the original lossless samples in the testing, if I understand correctly. Thus, it's not true ABX testing.