Index: doxia-core/src/main/java/org/apache/maven/doxia/parser/XhtmlBaseParser.java
===================================================================
--- doxia-core/src/main/java/org/apache/maven/doxia/parser/XhtmlBaseParser.java	(revision 628460)
+++ doxia-core/src/main/java/org/apache/maven/doxia/parser/XhtmlBaseParser.java	(working copy)
@@ -567,15 +567,26 @@
     {
         String text = parser.getText();
 
-        if ( StringUtils.isNotEmpty( text.trim() ) )
+        /*
+         * NOTE: Don't do any whitespace trimming here. Whitespace normalization has already been performed by the
+         * parser so any whitespace that makes it here is significant.
+         */ 
+        if ( StringUtils.isNotEmpty( text ) )
         {
-            // emit separate text events for different lines
-            String[] lines = StringUtils.split( text, EOL );
+            // Emit separate text events for different lines, e.g. the input "\nLine1\n\nLine2\n\n" should deliver the
+            // event sequence "\n", "Line1\n", "\n", "Line2\n", "\n". In other words, the concatenation of the text
+            // events must deliver the input sequence.
+            // (according to section 2.11 of the XML spec, parsers must normalize line breaks to "\n")
+            String[] lines = text.split( "\n", -1 );
 
-            for ( int i = 0; i < lines.length; i++ )
+            for ( int i = 0; i < lines.length - 1; i++ )
             {
-                sink.text( lines[i] );
+                sink.text( lines[i] + EOL );
             }
+            if ( lines[lines.length - 1].length() > 0 )
+            {
+                sink.text( lines[lines.length - 1] );
+            }
         }
     }