Earlier versions of IE set character encoding according to the language on the client when the document did not specify encoding. So at my computer with a Norwegian version of Microsoft Windows, documents without character encoding was parsed with ISO-8859-1 encoding.
But in IE7 this seems to have changed, at least in the current beta; the default character encoding is UTF-8. Now this can cause a major problem if you are not specifying a character encoding in the documents served to the client and at the same time using special characters such as Norwegian characters (ISO-8859-1). So on my test computer with IE7 installed all Norwegian characters show up scrambled because IE7 parse the content as unicode.
Solution: either specify character encoding according to the language used or use unicode in your pages (and make sure to specify UTF-8 as encoding as well).
And as final note it should be said that unicode is preferred regardless of the issue discussed here.