in server settings pane, click to remove deleted messages immediately.

Really? That is sooo strange. I enabled it, but what a weird place and wording to describe removing a spam email from the server. And why do we not get any control over which emails is does this with? Raises more questions than it answers, but thanks.

You may be misunderstanding. The setting I mentioned is to delete ALL from server. The only option I'm aware of for delete from server is for POP accounts.

Hm. I disabled that setting. Everywhere else I've looked, people say the filter action rule Delete Message From Server only works with IMAP. I did have two accounts set up in TBird before my computer's hard drive crashed—one using POP and one using IMAP—and I used to be able to enable this rule on every spam message that came in. Now it's not available as an option. I'll wait to see if anyone else has something to contribute here. Maybe there's something I'm just not getting.

You will be waiting for a very very long time is you are opinion shopping. This forum tends to have at most a dozen folk that actually answer questions here, most have been using Thunderbird for more than 10 years and I doubt any will have a similar recollection to yours. I certainly don't.

Now lets talk about email settings for on the server. If the account is IMAP then all mail is "on the server" by implication any action to delete the mail will delete the mail on the server.

POP mail can be left on the server, but there is no real connection between the local copy and the server copy. There is a setting in account setting to delete messages for the server when they are downloaded, or to leave them on the server until you delete them from Thunderbird. There is also a filter action to delete the message from the mail server. This action relies on the information stored in popstate.dat file of the account for that message. It is cumbersome and not the most reliable process. Corruption of the popstate.dat file can see thousands of mails left on the server downgraded again for no apparent reason. So leaving mail on the server is not a good choice, expecially if the space on the server is limited by the provider, but it is a popular one.

This image you posted is the filter menu of an IMAP mail account, which has by definition, no use at all of a delete from server, it is implied in the act of deleting.

When you create a filter the first thing you have to set, before you can create a filter is the account it applies to. Once that is set you get another dialog with the actual filter you are creating and the contents of the menu are context sensitive, you get the menu for the type of account the filters will be acting on. The following are the two distinct menu types for filter action. The first being the standard IMAP menu and the second that for a POP mail account with it's two additional menu entries "Delete from POP server" and "Fetch from POP server"

The reality is, regardless of your settings many servers ignore or reinterpret delete requests based on the commercial decisions of the provider. Outlook reinterprets deletion requests for POP to be a request to move the message to a folder called POP under the deleted items folder, but the message never actually goes away. Google relies on your choices on their web site for setting as to what to do when mail is downloaded in addition to what happens when the delete from the client is issued.

Email is not simplistic, despite a general idea in the community that it is "just email" and everyone knows how it works, the reality from where I am sitting is that email is quite a complex thing and basically no one knows how it works but assume, mostly incorrect things, and act outraged that it does not work how they assume it does.

I see from your post that you appear to have taken on the task of trying to manage spam with manual filters. I wish you luck with that as the reality is it is a self defeating task. Those sending spam are making millions from doing so, their highly paid experts will trump your efforts every time. I just rely on Thunderbird Bayesian junk filter and it serves me well. But my legal jurisdiction prohibits unsolicited commercial email with huge fines for business entities that breach those laws or fail to provide simple functional unsubscribe links.

I do however wish to offer one piece critical advice regarding manual spam filtering. If you are using a test based on the from address, do not mark the message as spam. Ever! The long term result will be you will probably make the self learning filter worse by poisoning the scoring database. Filtering takes no notice of the from address in its analysis of the message characteristics. You will instruct Thunderbird, via the filter to consider the text of the message as an indicator of spam and it is probably not the case. This is particularly contentious if you say use this filter as a form of blocklist. Messages from the blocked party wishing you happy birthday will become part of the learning database as an indicator of spam. Later all messages wishing you happy birthday may be blocked unless the sender is whitelisted because the phrase "happy birthday" is associated with spam.

Thank you for explaining and clarifying! This is so helpful. Of particular concern is your warning about manual spam filtering. Up to this point I have never missed anything important though I've doing doing it this way for years, but I understand the import of what you're saying. Sounds somewhat like the warning I give my clients about marking a spam message sent through their website as spam...they will never get mail from their website again! I just read https://blog.thunderbird.net/2024/09/thunderbird-and-spam/ and, between your comments and the information on the page, it appears that whether I use my filter rules or TBird's adaptive junk filter, it's going to train itself on the content of the messages. However, I've found that specific things like someone randomly asking if I want to sell my business continue to come through though I've added numerous email addresses sent to me for the same reason. Maybe I had the adaptive filter turned off in the old instance of TBird I had in my computer. So while I haven't experienced the difficulty you cautioned me about and I do benefit hugely from having those email addresses automically filtered whenever TBird checks for new emails, I do take your caution seriously. Just not 100% sure how to avoid it now, unless I turn off the adaptive Bayesian filter. Thoughts?