I am using sb_bnfilter.py version 1.1a3 on linux (actually CVS version from Aug 21), and am getting an increasing number of spam messages with 8-bit characters in messages labelled as 7-bit. These cause sb_bnfilter.py to give an error such as: UnicodeDecodeError: 'ascii' codec can't decode byte 0xb0 in position 1: ordinal not in range(128) and the message is not classified. I set the option replace_nonascii_chars: True in .spambayesrc, and some characters no longer caused problems (as they were replaced with ?). However any 8-bit character in the Subject: header still caused problems, also the character 0xb0 somewhere in messages. I used hexdump on the message to look for the reported 8-bit character. I could find the reported character in the message (or headers), except in the case of 0xb0, which did not appear to be anywhere in the message when dumped with hexdump. I have attached a message which reports the 0xb0 problem.