What is CVSS?

CVSS, the Common Vulnerability Scoring System, is an open framework from FIRST for describing how severe a software vulnerability is. It turns a set of technical characteristics into a number from 0.0 to 10.0 and a label from Low to Critical. The number ranks flaws against each other. It does not say how much a flaw matters to your own system.

Section Glossary Updated Measures severity

Quick facts

What it is
An open framework for communicating the characteristics and severity of software vulnerabilities (FIRST, CVSS v3.1 specification)
Maintained by
FIRST, at first.org/cvss
Newest version
4.0, first published 1 Nov 2023; the specification document is at revision 1.2, dated 18 Jun 2024 (FIRST, CVSS v4.0 specification)
Score range
0.0 to 10.0, mapped to None, Low, Medium, High and Critical, identically in 3.1 and 4.0
Metric groups
3.1: Base, Temporal, Environmental. 4.0: Base, Threat, Environmental, Supplemental
What it does not measure
Risk. The 3.1 user guide says CVSS “should not be used alone to assess risk” (FIRST, CVSS v3.1 user guide, section 2.1)

What a score is

An analyst looks at a vulnerability and picks a value for each metric in the specification: how it is reached, how hard it is, what it exposes. The specification’s equations turn those choices into one number. Two people who agree on the metric values get the same score, which is the point: a shared yardstick across vendors and advisories.

A CVE identifier and a CVSS score often sit side by side in an advisory, and they are different things. The identifier names the flaw. The score rates it.

The score also says nothing about whether an exploit exists. A flaw can score Critical with no known attack, and a Medium one can be in daily use. Part of the framework exists to record that difference, as the next section shows.

The metric groups

Base

The Base group describes the intrinsic qualities of a vulnerability, which the 3.1 user guide says are constant over time. In 3.1 the metrics are Attack Vector (Network, Adjacent, Local or Physical), Attack Complexity, Privileges Required, User Interaction, Scope, and the three impacts on Confidentiality, Integrity and Availability. Version 4.0 keeps most of these, adds Attack Requirements, and splits impact into the vulnerable system (VC, VI, VA) and subsequent systems (SC, SI, SA).

Temporal, or Threat in 4.0

This group captures what changes as time passes. In 3.1 it is called Temporal and has three metrics: Exploit Code Maturity, Remediation Level and Report Confidence. In 4.0 it is called Threat and has one, Exploit Maturity, with the values Attacked, Proof-of-Concept, Unreported and Not Defined. When no reliable threat intelligence exists, Not Defined is treated like Attacked, the worst case.

Environmental

Environmental metrics let the consumer adjust a score to their own deployment. In 3.1 these are the Confidentiality, Integrity and Availability Requirements plus a Modified version of each Base metric. This is the group that moves a score toward your situation and the only one that only you can fill in.

Supplemental (4.0 only)

Version 4.0 adds a group of extra context: Safety, Automatable, Provider Urgency, Recovery, Value Density and Vulnerability Response Effort. It describes the flaw further without being part of the Base, Threat or Environmental names.

Because a published score may include only some groups, 4.0 gives them labels: CVSS-B for Base alone, then CVSS-BT, CVSS-BE and CVSS-BTE as groups are added. On the question of which to use, FIRST says consumers “should enrich the Base metrics with Threat and Environmental metric values specific to their use” (v4.0 specification, introduction).

The 0.0 to 10.0 scale

Both specifications map the number to the same five qualitative ratings:

  • None: 0.0
  • Low: 0.1 to 3.9
  • Medium: 4.0 to 6.9
  • High: 7.0 to 8.9
  • Critical: 9.0 to 10.0

The bands are coarse on purpose. A 6.9 and a 7.0 are one tenth apart and land in different labels, so treat the band as a sorting aid and read the metrics behind it when it matters.

Reading a vector string

A score is usually shipped with a vector string, a compact record of the metric values behind it. It starts with the version, then lists metric and value pairs separated by slashes. The 3.1 specification gives this example:

CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:U/C:L/I:L/A:N

Read left to right: attack vector Network, attack complexity Low, privileges required High, no user interaction, scope Unchanged, low impact on confidentiality and integrity, none on availability. The 4.0 specification gives this one:

CVSS:4.0/AV:N/AC:L/AT:N/PR:H/UI:N/VC:L/VI:L/VA:N/SC:N/SI:N/SA:N/E:A

The last pair, E:A, is a Threat metric: Exploit Maturity set to Attacked. Its presence tells you this vector includes more than the Base group. The vector is the useful part of an advisory, because two flaws with the same number can differ in everything that matters to you: one needs a logged-in user, the other does not.

CVSS 3.1 vs 4.0

The version prefix in the vector tells you which specification produced the number. The differences, taken from the two specifications:

  • Threat replaces Temporal. Three metrics become one, Exploit Maturity.
  • Supplemental is new. Six metrics of context beyond the score groups.
  • Attack Requirements is new. It covers conditions in the deployment or execution environment that the attacker cannot control and that arise naturally, as opposed to security measures put in place to hinder attacks.
  • Scope is gone. The 4.0 specification says the idea is now captured by separating impact on the vulnerable system from impact on subsequent systems.
  • Named score types. CVSS-B, CVSS-BT, CVSS-BE and CVSS-BTE state which groups were used.

Compare like with like. A 3.1 number and a 4.0 number for the same flaw come from different metric sets, and the vector prefix is how you tell them apart.

Severity is not risk

FIRST is direct about this. The 3.1 user guide says CVSS “is designed to measure the severity of a vulnerability and should not be used alone to assess risk,” and that a comprehensive risk assessment should consider more factors than the Base score. Both specifications also list factors such as monetary loss as outside the scope of CVSS.

In practice the questions that turn severity into risk are yours to answer:

Do you run the component? A Critical flaw in software you never deployed costs nothing. Is it reachable? The same flaw behind a login on an internal network is a different problem from one on a public endpoint. Is anyone using it? The CISA Known Exploited Vulnerabilities catalog lists flaws with evidence of exploitation, and FIRST publishes EPSS, an estimate of the likelihood of exploitation. What does it hold? The impact of losing a marketing site and of losing a payments database are not the same, which is what the Environmental group is for.

A wide-reaching library flaw such as Log4Shell shows the pattern: the score opened the conversation, and whether you were affected depended on what you ran. An unfixed flaw may also arrive with no score at all, because rating takes time.

Where a WAF helps and where it does not

A web application firewall sees HTTP requests. It does not see your code, your dependency list or the CVSS score of anything in them, and it cannot repair a flaw. What it can sometimes do is refuse requests shaped like an attack on that flaw while the fix is being prepared, which is virtual patching. A score can help you decide where to look first. It does not tell a filter what to do.

That has limits worth stating. Many high-scoring flaws are not reachable through a request a filter can recognise: a local privilege issue, a flaw in a desktop client, a weakness in a build tool. A low-scoring flaw on a public form may be the one under attack. Neither outcome follows from the number.

When an advisory lands, start in monitor mode: read what the category matches look like in the log before you let anything refuse traffic, then move to block once you trust it. If the advisory turns out to describe something that already reached you, the steps in incident handling apply, whatever the score said.

Questions

What does CVSS stand for?

Common Vulnerability Scoring System. It is an open framework maintained by FIRST for communicating the characteristics and severity of software vulnerabilities.

What counts as a critical CVSS score?

In both version 3.1 and version 4.0 of the specification, Critical covers 9.0 to 10.0. High is 7.0 to 8.9, Medium is 4.0 to 6.9, Low is 0.1 to 3.9, and 0.0 is rated None.

Is CVSS the same as a CVE?

No. A CVE identifier names a vulnerability record; CVSS describes how severe the vulnerability is. An advisory often shows both, but one does not imply the other.

What changed between CVSS 3.1 and 4.0?

Version 4.0 renames the Temporal group to Threat and reduces it to one metric, Exploit Maturity. It adds a Supplemental group and an Attack Requirements metric, and replaces Scope with separate impact metrics for the vulnerable system and subsequent systems.

Does a high CVSS score mean my site is at risk?

Not by itself. FIRST states that CVSS is designed to measure the severity of a vulnerability and should not be used alone to assess risk. Whether the flawed component is reachable on your site, and whether anyone is exploiting it, decide how much it matters to you.