gregor, I think its not well formedness in that start tags not equal to end tags or anything...

In the code that I posted below, I set error context, and I get the place where the error is supposed to occur.. and this is inside the CDATA section, or in the embedded text...

I also don't think its (the file) is corrupt, because I extract it from a zip, and the unzip gives no errors..Unfortunately, the header only says <xml version="1.0"> no news of what encoding it is...

The only option that I see is to do a search and replace to strip out the control characters before I parse.. I really don't see any other choice, but I'm willing to listen to anyone who tells me this is too drastic :)

Thanks