I probably wasn't precise enough in my first post: All rips were done with the same drive.
I have now checked the files with EAC's comparison tool, and it turns out the audio contents of the files were indeed identical. This is a first for me! You must be right that this has to do with tagging of wav files in MC 17, which has not been done in previous versions.
Doh... I don't know why this has been implemented. Can I turn it off and make it work like before?
A part of my work flow has always been to rip first to wav, create an md5 checksum file for the wav files, then convert to lossless. It has been very useful to me, and I don't know any other way to verify later that a file is what I think it is. There is always the danger of accidentally converting via some lossy format, and still believing the file to be lossless.
The checksum test has always been very handy. I can convert a file to wav, check it against my original checksum, and know that it is indeed identical to the wav file I ripped. Do any of you have any other method that you can recommend?