What Is a Spam Score?
TL;DR
A spam score is a number a content filter assigns to a message by adding the weight of every rule the message triggers. SpamAssassin is the reference implementation: its default threshold is 5.0, so a message scoring 5.0 or above is marked as spam and 4.9 passes. Positive weights count against the message, negative weights in favour.
How is a spam score calculated?
The filter runs hundreds of independent tests over the headers, the body and the DNS records behind them, and each test that fires contributes its own weight. The weights are summed, and the total is compared with the configured threshold, which SpamAssassin calls required_score and ships at 5.0.
The threshold is a policy choice rather than a law. The SpamAssassin documentation describes 5.0 as aggressive and suited to a single user, and suggests a provider filtering for many mailboxes set something more conservative, so the same message can be spam on one server and clean on the next.
Does a good spam score mean you reach the inbox?
No, and this is the most common misreading of a spam-check report. Google and Microsoft do not run SpamAssassin, and their filters weigh sender reputation and recipient engagement above content, so a score of zero says only that a rule-based content filter found nothing to object to.
Read it as a hygiene check. It catches real faults worth fixing: a missing or misaligned authentication result, a body that is only an image, a link shortener, a blocklisted sending IP. It does not forecast where a large provider will file the message.
Frequently asked questions
Below the threshold configured on the servers your recipients use, which you cannot know, so aim as close to zero as the message allows. Under SpamAssassin defaults anything from 5.0 up is marked as spam, and a score creeping toward 3 or 4 usually means several small faults have accumulated rather than one large one.