Index: src/main/java/com/xpn/xwiki/wysiwyg/server/plugin/importer/XWikiImportService.java
===================================================================
--- src/main/java/com/xpn/xwiki/wysiwyg/server/plugin/importer/XWikiImportService.java	(revision 25069)
+++ src/main/java/com/xpn/xwiki/wysiwyg/server/plugin/importer/XWikiImportService.java	(working copy)
@@ -83,10 +83,15 @@
     public String cleanOfficeHTML(String htmlPaste, String cleanerHint, Map<String, String> cleaningParams)
     {
         try {
+            // First removes the <?xml:namespace .../> declarations/instructions that ms word is wrongly adding and IE
+            // is wrongly preserving upon serialization of the pasted HTML.
+            // Remove all sequences starting with <?xml:namespace and ending with /> provided there is no other <
+            // between them (this would make the closing sequence belong to another tag)
+            String cleanedPastedHTML = htmlPaste.replaceAll("\\Q<?xml:namespace\\E[^<]*\\Q/>\\E", "");
             HTMLCleaner cleaner = componentManager.lookup(HTMLCleaner.class, cleanerHint);
             HTMLCleanerConfiguration configuration = cleaner.getDefaultConfiguration();
             configuration.setParameters(cleaningParams);
-            Document cleanedDocument = cleaner.clean(new StringReader(htmlPaste), configuration);
+            Document cleanedDocument = cleaner.clean(new StringReader(cleanedPastedHTML), configuration);
             HTMLUtils.stripHTMLEnvelope(cleanedDocument);
             return HTMLUtils.toString(cleanedDocument);
         } catch (Exception e) {