Hi > This seems to be true for me also. Strange that I never noticed before. > Am I right in thinking that patch > > id:1433843467-32398-1-git-send-email-markwalters1009@gmail.com > > is directed to fix this? Yes that is correct. What happened is that the tree code relied on the message-pane (a show buffer) to mark the message read, but updated its own display. After the unread handling changes the show buffer only marks a message read if it is the active buffer (so its post-command hook gets called), so it no longer removes the unread tag. Best wishes Mark (PS this message shows the deficiencies in our emacs reply handling: it doesn't include the rfc822 part!)