SpamSieve assigns each message a score between 0 and 100, and it considers messages with scores 50 or greater to be spam. The farther the score is away from the middle (50), the more confident SpamSieve is about whether the message is spam. Put another way, if you are trying to double-check SpamSieve’s work, you should focus your attention on the messages with scores closer to 50. If SpamSieve has made any mistakes, it is more likely to be with messages scored at 70 or 30 than with those scored at 95 or 5.
Spam messages that SpamSieve thought were good (false negatives) are generally easy for humans to spot. They tend to stick out amongst lots of good messages in the inbox.
Good messages that SpamSieve thought were spam (false positives) are harder to spot. They can be buried amongst piles of messages in the junk mailbox. To help you find these messages, SpamSieve can notify you about uncertain spam messages, i.e. the messages in the junk mailbox that you should focus your attention on.
The pop-up menu lets you control the range of scores that SpamSieve considers to be uncertain. For example, if you set it to 75 (the default), messages that SpamSieve thought were spam but which have scores below 75 will be considered uncertain.
By making it easier to spot uncertain messages, these mechanisms help you to efficiently verify whether SpamSieve has filtered your mail correctly.
Uncertain messages are considered to be spam, unless you tell SpamSieve otherwise, so you should not train SpamSieve that they are spam.
See also: Move it to the Trash if the spam score is at least.