--- mailman-2.1.8/Mailman/Handlers/SMTPDirect.py.sender-header	2005-12-30 13:50:08.000000000 -0500
+++ mailman-2.1.8/Mailman/Handlers/SMTPDirect.py	2006-05-02 13:45:21.000000000 -0400
@@ -347,17 +347,23 @@
 
 
 def bulkdeliver(mlist, msg, msgdata, envsender, failures, conn):
-    # Do some final cleanup of the message header.  Start by blowing away
-    # any the Sender: and Errors-To: headers so remote MTAs won't be
-    # tempted to delivery bounces there instead of our envelope sender
-    #
-    # BAW An interpretation of RFCs 2822 and 2076 could argue for not touching
-    # the Sender header at all.  Brad Knowles points out that MTAs tend to
-    # wipe existing Return-Path headers, and old MTAs may still honor
-    # Errors-To while new ones will at worst ignore the header.
-    del msg['sender']
+    # Do some final cleanup of the message header.  Start by blowing away any
+    # Errors-To: headers so remote MTAs won't be tempted to delivery bounces
+    # there instead of our envelope sender.
+    # 
+    # Previously, we also blew away any Sender: headers, and added a new
+    # Sender: header with our envelope sender.  However, this behavior seems to
+    # cause more problems than it solves, because various MUAs will include the
+    # Sender address in a reply-to-all, which is not only confusing to
+    # subscribers, but can actually disable/unsubscribe them from lists,
+    # depending on how often they accidentally reply to it.
+    # 
+    # The drawback of not touching the Sender: header is that some MTAs might
+    # still send bounces to it, so by not trapping it, we can miss bounces.
+    # (Or worse, MTAs might send bounces to the From: address if they can't
+    # find a Sender: header.)  But at least for now, we're going to take our
+    # chances with leaving the Sender: header alone.
     del msg['errors-to']
-    msg['Sender'] = envsender
     msg['Errors-To'] = envsender
     # Get the plain, flattened text of the message, sans unixfrom
     msgtext = msg.as_string()