Note: I'm from Germany, so i'm familiar with the problem of special chars. Not only in having correct encodings in JRXML files. And my experience with Eclipse related to encoding stuff is: it's ugly, it's intransparent. Never use Eclipse to change or repair encodings in texts. (That's not Eclipse specific, the most IDEs seem to be the wrong tools for this task) Use a relativ simple editor (i use ConText). Open the file in the editor and remove the CDATA sections and texts in the static text boxes. Save the file and try to compile it. If it works, go back to the editor and add the texts again. ConText stores the file in ISO-8859-1 by default, so take care of setting <?xml version='1.0' encoding='ISO-8859-1' ?> Compile again, if it works, be happy and be careful, if you touch the JRXML in the future again. If it does not compile with empty text boxes, then you may have unusal chars somewhere in the file. You could enable ConText feature to show you unprintable chars in the file and look over it. But i guess (seeing the file size) this would require pratically more time then setting up a new report template from scratch.