# The NVD is the reference database for CVE records, and since April 2026 it enriches only a prioritized part of them

SCORING SYSTEM, read 10 Nov 2026

**The short answer.** The National Vulnerability Database, run by NIST, adds severity scores, product lists and weakness data to CVE records. On 2026-04-15 NIST moved to risk-based enrichment: KEV entries first, with a one business day goal. CVE submissions rose 263 percent from 2020 to 2025, and 130 of 318 recent KEV records carry an NVD-sourced score.

- **Maintained by:** NIST, Information Technology Laboratory
- **Scale:** Not a scale; carries CVSS (v2, v3.x, v4.0) from NVD or CNAs, plus CWE, CPE and, since 2026-06-17, CISA SSVC data
- **Current version:** API 2.0; SSVC schema v2.0.3; CVE Affected v1.0
- **Updated:** Continuously; announcements on the NVD news page
- **Answers:** What the CVE record says about affected products, weakness and scored severity

## What it is

NIST states on the NVD home page that it maintains the National Vulnerability Database as a repository of information on software and hardware flaws that can compromise computer security. The database holds the CVE record, the NVD analysis added to it, and fields from other sources such as the CNA that assigned the CVE and CISA.

Every statistic on this site that mentions a CVE's published date, its CVSS value or its source comes from NVD API 2.0 (services.nvd.nist.gov/rest/json/cves/2.0). The published date in that record is the start of the lag clock in the KEV Lag Ledger.

The NVD is also the place where CISA data meets the CVE record. The 2026-05-28 notice says NVD already included KEV catalog data and other enrichment present in the CVE record, and would add SSVC data and affected-product information in the feed and API. That makes NVD one integration point for three of the signals on this site: CVSS, KEV and SSVC.

## The April 2026 change

On 2026-04-15 NIST announced a risk-based model driven by record CVE growth. Its figures: CVE submissions rose 263 percent between 2020 and 2025, submissions in the first three months of 2026 ran nearly one-third above the same period of 2025, and NIST enriched nearly 42,000 CVEs in 2025, 45 percent more than in any earlier year.

From 2026-04-15 NIST prioritizes three groups for enrichment: CVEs in the CISA KEV catalog, with a goal of one business day from receipt; CVEs for software used within the federal government; and CVEs for critical software as defined by Executive Order 14028. Other CVEs are still added but categorized Lowest Priority, not scheduled for immediate enrichment. Users can ask NVD to enrich a specific CVE by email.

The notice states that these criteria may not catch every potentially high-impact CVE, which is why NIST offers the email request route. A vulnerability management lead should expect some CVEs with a high EPSS score and no NVD severity, and plan to take the CNA score or the CISA data in that case.

## Severity scores: who supplies them

The same notice states that NIST will no longer routinely provide a separate severity score for CVEs where the CNA has already provided one. Users can request one for a specific CVE.

The ledger shows how mixed the source already is. Of 318 KEV additions, 130 (40.9 percent) take their CVSS score from nvd@nist.gov; the rest come from CNAs, among them secure@microsoft.com (28 entries) and psirt@cisco.com (16). A program that treats an NVD score as one authority's number is mixing scores from many assessors. Store the source field.

The scope of the 2026-04-28 correction is worth noting. NIST said only the calculated numerical v4.0 scores were affected, that scores had been too high in about 4,500 records and too low in fewer than 30, and that API users who sync with lastModStartDate and lastModEndDate would pick up corrections automatically; users who do not would need to refresh.

The ledger takes one more thing from the NVD record: the CVSS version. Of the 318 scores, 313 are v3.1 and 5 are v3.0, which reflects what the NVD and the CNAs had published for these CVEs by the read date. A program planning for v4.0 should check what share of its own open findings carry a v4.0 vector from any source before changing its policy.

| Date | NVD notice | Effect on a program |
| --- | --- | --- |
| 2026-04-15 | Risk-based enrichment; no routine NIST score where a CNA scored | Unenriched CVEs may have no product list; severity may come from the CNA |
| 2026-04-28 | About 4,500 v4.0 records corrected (19 percent of v4.0 records) | Resync scores using lastModStartDate and lastModEndDate |
| 2026-05-28 | Announcement of SSVC and affected data in CVE feed and API | Plan for larger payloads |
| 2026-06-17 | SSVC (ssvc-v2.0.3) and CVE Affected v1.0 deployed; about 95 percent of records touched | Larger CVE-Modified feed for eight days |
| 2026-08-26 | Audit history entries link to the CVE on GitHub instead of embedding affected data | Change feed parsers |

## What the KEV Lag Ledger measured against it

The ledger matched all 318 KEV additions in its window to an NVD record. For each it records the NVD published date and computes days from publication to listing. The median is 10.5 days, 45.3 percent fall within 7 days, and 17.3 percent (55) were published more than 365 days before listing.

Because NVD publication is the clock start, the lag is not the time from first exploitation. Eight additions have negative values: CISA listed them before the NVD published the record. The metric describes how long a CVE sat visible in the NVD before CISA confirmed exploitation, which is what a scanner feed could have shown.

The 55 additions published more than a year before listing are a reminder that NVD age and risk are separate. An old record can be fully enriched, carry a Critical score for years and still reach KEV only when attackers take it up. NVD completeness does not predict exploitation. The same holds in reverse: an unenriched record is not a low-risk record, and the 2026-04-15 notice says so, calling the criteria a way to focus on the CVEs with the greatest potential for widespread impact, not a ruling on the rest.

## Using it in a ranking

Use the NVD for identity and affected products, and treat its CVSS value as one assessor's. Fields to store with each score: source, version, vector string, and the lastModified timestamp. After 2026-04-15, a CVE with no NVD product list is a CVE NVD has not yet prioritized, not a CVE with no affected product.

KEV entries are the best-enriched records by policy. For everything else, expect a delay, and let the CNA's data, EPSS and the CISA SSVC fields carry the ranking until enrichment arrives.

A sensible store keeps a raw copy of the NVD JSON per CVE with the fetch date, so any later question about what the NVD said on a given day has an answer.

## Comparisons with this system

- [CVSS vs EPSS vs KEV](/compare/cvss-vs-epss-vs-kev/)

## Sources

- NVD news and updates (2026-04-15, 04-28, 05-28, 06-17, 08-26): https://nvd.nist.gov/general/news (read 10 Nov 2026)
- CISA, Vulnrichment README: https://github.com/cisagov/vulnrichment (read 10 Nov 2026)
- FIRST, Using EPSS: https://www.first.org/epss/using-epss (read 10 Nov 2026)

Last reviewed 10 Nov 2026. Canonical: https://epssvscvss.com/scores/nvd/
