I just began working with the METS schema and I think I've discovered a couple problems that I want to report:
1) In the enumerations for the MDTYPE attribute, "VRA" occurs twice.
2) The mdWrap element contains overlapping particles ("binData" and "any"), which makes the content model ambiguous. What this means: In the METS schema, it says we can either have "any" (wildcard...any element) or "binData". Since binData is in the set of all possible elements, these choices overlap. You would be saying the same thing if you simply said "any" and didn't make it a choice. More info on the constraint this violates is at http://www.w3.org/TR/2001/PR-xmlschema-1-20010316/#cos-nonambig and http://www.w3.org/TR/2001/PR-xmlschema-1-20010316/#non-ambig
These types of problems won't be caught by validating the METS schema against the XML Schema schema -- The tool I used to find these is at http://www.alphaworks.ibm.com/tech/xmlsqc, and does more thorough validity checking on schemas. I've just recently discovered it and would recommend it to anyone doing development in this area.