Historically, fossil dinosaur bones have been considered to be entirely mineralized, that is, that they are faithful replicas of the bones but don't contain bone per se. However, in 2005, the laboratory of Mary Schweitzer fortuitously observed what looked to be soft-tissue blood vessels and even blood cells in a Tyrannosaurus leg bone. Schweitzer's lab has since worked on recovering and identifying proteins , specifically collagen, in fossilized dinosaur bones. However, many in the field find it hard to believe that 65 million-year-old proteins would survive the fossilization process, and I would say a substantial
Schweitzer has more recently collaborated with collagen expert Joseph Orgel in a very recent paper in the open-access jounal PLOS One, which at least offers an explanation for which protein fragments their detection methods are registering. Collagen, the major protein in bones, is wound into incredibly tough three-stranded helical fibrils, and would be as good a candidate as any to survive fossilization processes. And there are patterns to the portions of collagen detected by their experiments. Collagen fragments detected in analysis of two different fossil dinosaur bones seemed to derive from the same place in the collagen fibril, consistent with the idea that those regions are better preserved than others. Secondly, the fragments from both species correspond to sections of collagen known to lie in the interior of the triple helix. By being shielded here, the authors hypothesize, these fragments were more likely to be preserved in sufficient amounts to be detected.
I would say that while these two factors could explain detection of fossil collagen, the same shielding and similarity effects would also apply to contaminating avian or even reptilian collagen. And it wouldn't take much modern protein contamination to drown out a 65 million year-old signal. So for myself, I still find this a tantalizing possibility, but still scientifically ambiguous.