Your suggestion worked perfectly for a few days. Then the length of the block of binary data grew. The BOM character was no longer in the same place. When I seek to the old location, the read fails. Sure, I can fix the seek position and it will work for a while again. I still think that I need Choroba's original approach. Read the file in binary, extract the XML with a regular expression, convert the XML string to ASCII and then parse. I am not able to do the conversion step.