Training - Configure how the filter learns as it processes messages

DSPAM should train: On every new message scanned by the filter Only when the filter makes a mistake Only with new data or if the filter makes a mistake

When I train DSPAM, I prefer: To forward my spams (signature appears in message body) To bounce my spams (signature appears in message headers)

Filter sensitivity during the training period:
Catch SPAM (More in Quarantine) | | Assume Good (Fewer in Quarantine)

Message Handling - Configure how SPAM is handled

When a SPAM message is identified:

Quarantine the message
Tag the Subject header with
Deliver the message normally with a X-DSPAM-Result header

Features - Tuning SPAM filtering

$OPTION$ Enable noise reduction, which usually improves filtering accuracy
Enable automatic whitelisting to record frequent correspondence
Add the factoring tokens in each email into the message's full headers