Skip to content
StratQuad

Common Vulnerability Scoring System (CVSS)

4 min read

FIRST's open standard for describing the characteristics and severity of a software vulnerability as a vector and a score from 0.0 to 10.0.

At a glance

Owner
Forum of Incident Response and Security Teams (FIRST), CVSS Special Interest Group
Version
CVSS v4.0
Published
1 Nov 2023
Applies to
Vulnerability publishers, vendors and every organisation prioritising remediation
Mandatory
Voluntary standard; referenced by schemes such as Cyber Essentials and many contracts
Assessed by
Scores are assigned by the vendor, a CVE Numbering Authority or NVD, and adjusted by the consumer
Cost model
Free to use

Key points

  • CVSS measures severity, not risk. A Base score describes the vulnerability in isolation; it says nothing about whether it is being exploited or whether it matters to you.
  • CVSS v4.0, published on 1 November 2023, has four metric groups: Base, Threat, Environmental and Supplemental.
  • v4.0 replaces Scope with separate impacts on the vulnerable system and subsequent systems, adds Attack Requirements, and reduces the old Temporal group to a single Exploit Maturity metric.
  • Scores should be reported with their nomenclature: CVSS-B (Base only), CVSS-BT, CVSS-BE or CVSS-BTE.
  • CVSS v3.1 remains widely published alongside v4.0, and many policies, including Cyber Essentials, still reference v3 scores.

What it is

CVSS turns a set of observable characteristics of a vulnerability into a vector string and a numeric score. The vector is the useful part: it records how the vulnerability is reached, what conditions must hold, and what is affected. The score is a summary of that vector.

FIRST maintains the standard. Vendors, CVE Numbering Authorities and the US National Vulnerability Database (NVD) publish Base scores; consumers are expected to refine them with Threat and Environmental metrics.

Metric groups in v4.0

  • Base: Exploitability
    • AVAttack Vector

      Network, Adjacent, Local, Physical

    • ACAttack Complexity

      Low, High

    • ATAttack Requirements

      None, Present

    • PRPrivileges Required

      None, Low, High

    • UIUser Interaction

      None, Passive, Active

  • Base: Impact
    • VC, VI, VAVulnerable System Confidentiality, Integrity, Availability

      High, Low, None

    • SC, SI, SASubsequent System Confidentiality, Integrity, Availability

      High, Low, None

  • Threat
    • EExploit Maturity

      Attacked, Proof of Concept, Unreported, Not Defined

  • Environmental
    • CR, IR, ARConfidentiality, Integrity and Availability Requirements

      High, Medium, Low, Not Defined

    • Modified Base metrics (MAV to MSA), letting the consumer reflect their own deployment; MSI and MSA also accept Safety
  • Supplemental

    (Informational, does not change the score)

    • Safety (S), Automatable (AU), Recovery (R), Value Density (V), Vulnerability Response Effort (RE), Provider Urgency (U)

A v4.0 vector starts with CVSS:4.0/ and lists metrics in a fixed order. All eleven Base metrics are mandatory; omitted Threat, Environmental and Supplemental metrics are treated as Not Defined. For example:

Code
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

Qualitative severity

RatingScore
None0.0
Low0.1 to 3.9
Medium4.0 to 6.9
High7.0 to 8.9
Critical9.0 to 10.0

FIRST describes the qualitative scale as optional. The bands are the same in v3.1 and v4.0.

How v4.0 calculates a score

v3.1 used a formula. v4.0 does not. Vectors are grouped into MacroVectors using six equivalence sets, each covering related metrics (for example EQ1 covers AV, PR and UI). Each MacroVector has a score set by expert review, held in a lookup table. Vectors within a MacroVector are scored by interpolating between it and the next lower MacroVector.

The practical consequence: v4.0 scores cannot be reproduced by hand. Use the FIRST reference calculator or a library that implements its lookup table.

What changed from v3.1

Changes from CVSS v3.1 to CVSS v4.0

  • Scope removed. Replaced by separate impact metrics for the vulnerable system (VC, VI, VA) and subsequent systems (SC, SI, SA).
  • Attack Requirements (AT) added. Captures deployment conditions outside the attacker's control, such as a race condition that must be won or a need to be on the network path, which v3.1 folded into Attack Complexity.
  • User Interaction refined. Required becomes Passive (limited, involuntary interaction, such as visiting a malicious page) or Active (specific, conscious action, such as importing a file in a particular way or accepting a security warning).
  • Temporal becomes Threat. Only Exploit Maturity remains; Remediation Level and Report Confidence are removed.
  • Supplemental metrics added. Extra context, such as whether exploitation is automatable or affects safety, without altering the score.
  • Nomenclature introduced. CVSS-B, CVSS-BT, CVSS-BE and CVSS-BTE make clear which groups a score includes.
  • Scoring by lookup instead of formula, as above.

Using it well

Practitioner note

Treat the published Base score as the vendor's description of the worst case, then apply two adjustments before it drives a remediation deadline. First, Threat: if the vulnerability is in the CISA KEV catalogue or has a high EPSS probability, it is being, or is likely to be, exploited. Second, Environment: an internet facing asset holding sensitive data deserves a higher priority than the same software on an isolated test server. A programme that patches by Base score alone spends most of its effort on vulnerabilities that are never exploited.

Note

Cyber Essentials uses CVSS v3 Base scores in its 14 day update rule: an update must be applied within 14 days if the vendor describes the vulnerabilities it fixes as critical or high risk, if they have a CVSS v3 base score of 7 or above, or if the vendor gives no details of their severity. Check which version a policy refers to before applying v4.0 scores to it.

Common mistakes

Common nonconformity

  • Reporting "CVSS 9.8" with no version or vector, so readers cannot tell what was scored.
  • Comparing v3.1 and v4.0 scores directly. The scales share bands but not calculations, and the same vulnerability can score differently.
  • Treating severity as likelihood. A critical Base score on an unexploited flaw in an unused component may warrant less urgency than a medium score on an actively exploited internet facing service.
  • Never recording Environmental adjustments, so prioritisation cannot be explained to an auditor afterwards.

Sources

Suggest a correction