As others have said, the byte-order mark for UTF-8 is not necessary, and I would even recommend against it (for the same reason that the W3C does).
Unfortunately getting rid of it may be tricky, depending on which text editor you use. If the text editor supports UTF-8 (which most do now) then you won't even see the byte-order mark in the file when you open it. You would have to rely on there being a menu item to choose between character encodings, one of which may be "UTF-8 (no byte-order mark)" or even just "UTF-8".
As a quick and dirty hack you could try opening it in a program that does NOT support UTF-8 and then you'll see the byte order mark as three strange characters and you'll be able to just delete them. Take care that it doesn't mess up any special characters elsewhere in the page though. And I can't suggest a non-UTF-8 aware application off the top of my head, but most full featured text editors like Notepad++ or PSPad will allow you to switch between UTF-8 and other modes.
The purpose of the byte-order mark is to make sure that your computer system is not reading every sequence of 2 bytes in the wrong order. However, that is not relevant to UTF-8, because UTF-8 is encoded to a stream of single bytes and thus it has no byte-order issues. The byte-order mark is therefore useless (except as a "hint" that the document uses UTF-8 encoding, which is unnecessary).
Edit: oops, missed the fact that the OP's problem has already been solved. Oh well, looks like I wasn't the only one
Technically, that's not completely correct - while some implementations may determine the default character encoding of an HTML and XML document differently in the absense of any indication, normally you would indicate the character encoding somewhere in the document. It is indeed legitimate to have ISO-8859-1 or even CP-1251 (microsoft) in an XML document as with HTML. The statement "XML 1.0 requires unicode" is misleading here. Both HTML and XML always represent characters internally by their Unicode code point, and all XML implementations need to be able to support UTF-8 and UTF-16, but that doesn't mean the document needs to use either of those encodings where the intended implementation supports others.