Growl giving nonsense as well as accurate reports

SpamSieve is doing beautifully for me. But Growl is a little peculiar. It does pick up an accurate count on incoming mail, but it also issues other bulletins that make no sense. “23 good messages”, for instance. I see this, check my Mail program, and there may be, say 21 messages in the Inbox that have already been vetted when they came in. I’ve noticed that if I check a mailbox On My Mac to read something I received a while ago, it sometimes sends me notice of the number of ‘good messages’ in that mailbox.
This is not a problem for me. It is nothing like getting the spam I was getting before I discovered SpamSieve. But it would be nice to get it to not do this.
I repeat – the accurate information is getting to me through Growl; there is also extra information that has no meaning for me.

SpamSieve 2.9.6, OS X 10.7.5, Mail 5.3, Growl 1.2.2

I don’t understand. Which Growl notification do you think is accurate and which one do you think is not?

The Growl notifications have nothing to do with which mailbox you’re looking at or which messages you’re reading. Please see this description of how it’s designed to work.

You can also turn off any notifications that you don’t wish to see.

which is the good growl?

Good Growl: “3 Good Messages”, and I look and there are three new messages.
Bad Growl: 10 or 15 or 20 minutes, 3 hours, later, no discernible pattern, “27 Good Messages”, I look, nothing new.
Or: “1 Good Message” , I look, nothing new.
Or: “4 Good Messages”, I look, nothing new.

OK, so it’s designed to reset if I train a message, but many times when these random announcements arrive, I have not trained it, nor restarted Mail, nothing.

It’s difficult for me to understand how I would turn off the Bad Growl messages without turning off the ones that are useful.

I don’t understand what you don’t understand.

I initially thought you were referring to certain types of notifications that were unwanted. Now I see that you’re referring to a single type (“Good Message Count”) but that you sometimes think it’s reporting incorrect numbers. I suggest that you look in the log to see which new good messages contributed to the count.