I'd say it probably is a no-no. At this point I'm sure you can hear the audible difference when A/Bing an MP3 (or any compressed format) vs. WAV/AIFF (or any uncompressed format). If you referencing a track that's been squashed, you may end up taming your mix based on the "missing" frequencies of said reference track–resulting in your mix not living up to its full potential.
I also keep Logic's Spectrum analyzer permanently on my 2nd monitor, it really helps when A/Bing your track vs. reference track, as you can see where certain frequencies are emphasized over others such as kick and snare peaks for example.
Ah balls i figured it would be a bad thing to do. Will have to buy some WAVs in future then. When referencing, ive found the analyser to be the best tool to use too. Doing hi and low passes across the spectrum too to see where stuff starts and ends etc.
Hiya, glad this is well received. As it is a critical listening task it is fundamental to use a high quality audio source. It may not be the end of the world if you are 320kbps. As we all know extended top can go AWOL in some 128 files (steep LP filtering in some versions of MP3 codec) along with that swirling mushy blurred upper mid and highs. Bass tends to lose solidity and definition as well. Ultimately you can probably glean the basic tonal content from an mp3 but the clearer the reference the less ambiguity, and less ambiguity is where good sound is at.