In today’s Washington Post, Michael Tedeschi has a column on lossless audio formats. While it has some good information, it starts with a crucial error:
You may not have realized that the song playing through the earbuds of your MP3 player is actually a lower-quality, digitally compressed version of the original recording.
But try playing that same MP3 file on a high-end home stereo system and you’ll likely hear some hisses and snaps and pops in the music — even if you bought it from a download service such as iTunes or ripped it from your own copy of the original CD.
Sorry, but that’s not correct. A compressed audio file sounds worse than the original because big chunks of bandwidth have been chopped off. The highs and lows have been removed to save space. Update: As Dan notes in the comments, there’s more to the story than my original explanation would suggest: “Mp3s sound bad because frequency data has been altered and removed throughout the entire audible spectrum. The cleverer versions of the algorithm will alter the critical midrange, where one’s hearing is most acute, less than the extremes of the spectrum, but, in general, chunks of highs and lows aren’t just getting “chopped off.” The entire spectrum is being simplified (gutted, if you will).”
Clicks and pops can occur even if you use a lossless format like FLAC or Shorten or Windows Media Lossless or Apple Lossless. That usually happens during the ripping process. If there’s a scratch or other flaw on the CD, or if your CD drive is less than perfect, or if your CPU can’t keep up with the data being extracted, the result is a digital copy filled with errors.
In fact, every time you rip a CD using a consumer-grade tool, you’re probably getting less-than-perfect copies. If you’re just making copies for your portable music player, this might be no big deal, but if you’re an audiophile exchanging copies of live performances, it’s hugely important.
The solution? Try Exact Audio Copy. This postcard-ware program is an audiophile’s dream. Used in secure mode, it makes copies that are as close to perfect as you can get.
Highly recommended.
Ed Bott 
“A compressed audio file sounds worse than the original because big chunks of bandwidth have been chopped off. The highs and lows have been removed to save space.”
Sorry, but that’s not really correct, either. Sure, it’s standard to chop the top 3kHz or so of range off the top of a CD quality track (and I think this is done to make the math easier, not so much to save space), but most of that frequency range is inaudible to adult humans, and probably doesn’t even make it out of your speakers, anyway. Sometimes, if something is going to be heavily compressed, it will get a whole swath of high end removed (like the top HALF of the audible spectrum), but this isn’t what’s done with the vast majority of commercially available mp3 files.
Mp3s sound bad is because frequency data has been altered and removed throughout the entire audible spectrum. The cleverer versions of the algorithm will alter the critical midrange, where one’s hearing is most acute, less than the extremes of the spectrum, but, in general, chunks of highs and lows aren’t just getting “chopped off.” The entire spectrum is being simplified (gutted, if you will).