There's going to be a really cool symposium at the AAAS meeting in St. Louis about analysis of protein recovered from fossils. Some proteins and DNA can survive the fossilization process and be preserved for amazing periods. Meanwhile, methods for recovering and analyzing tiny amounts of these substances continue to get better and better.
One speaker at the symposium, Peggy Ostrom, is using mass spectrometry to sequence proteins (probably osteocalcin ) from bone powder half a million years old.
For an excellent article about the challenges still facing ancient DNA analysis, go here . The caveats and opportunities listed there are probably going to apply to paleoproteomics as well. For example, protein analysis of osteocalcin should be pretty hard to troubleshoot, because the human protein is going to be almost identical to what you'd expect from any mammalian fossil.
UPDATE: Some more details on the symposium at afarensis