" Every spam filter is forced to weigh the relative cost of a false positive (good mail quarantined as spam) and a false negative (spam mail let through to the inbox). Whilst a spam mail in the inbox has an immediate small cost, a false positive has a longer term larger cost. If the false positive represented potential business then even a single false positive can vastly outweigh the cost of the spam surrounding it. As spam volumes increase massively, manually searching a spam folder for messages has become impossible. But the cost of a false positive has remained the same. This paper reports on a project called 'False Positive Finder' that operates once a spam filter has acted to examine spam held in a quarantine for potential false positives, presenting the end-user with a small number of messages to examine from the sea of spam. False Positive Finder uses a continuous scoring technique to look for false positives not just in recently arrived messages, but in historic messages that may be hours, days or weeks old. "