Skip to content

Vulnerability Statistics 2026: CVE Volume, Exploitation Speed and Patch Times

Vulnerability statistics for 2026 counted from CVE.org, NVD and CISA KEV data, plus Verizon DBIR, Mandiant M-Trends, Google and VulnCheck exploitation research.

Mark KissBy , Founder & CEOUpdated 16 min read

Penetration Testing ServicesExternal network from $4,200, web app from $5,200Fixed scope, free retest, no hourly billing.

Get my fixed price

This page tracks how many vulnerabilities are published, how severe they are, how many get exploited, how fast, and how long organizations take to fix them. Where official data exists we counted it ourselves, on November 16, 2026: CVE totals from CVE.org, severity and backlog figures from the NVD’s JSON feeds, and exploitation figures from the CISA Known Exploited Vulnerabilities (KEV) catalog file. Those figures are marked “Invadel count”. The rest comes from Verizon, Mandiant, Google Threat Intelligence Group, VulnCheck, OWASP and other primary sources, linked beside each figure.

How to cite: link to this page or to the primary source beside each figure. Survey figures carry their sample size.

Key numbers at a glance

  • 48,244 CVE records were published in 2025, up 20% from 40,077 in 2024. (CVE.org Metrics, 2025)
  • 36,962 CVE records were published in the third quarter of 2026 alone, more than three times the 11,738 of the third quarter of 2025. (CVE.org Metrics, 2026)
  • 1,781 vulnerabilities are in the CISA KEV catalog, and 297 were added in 2026 through November 13, already more than in any full year since 2022. (Invadel count of the CISA KEV catalog, version 2026.11.13)
  • 31% of breaches with a known initial access vector started with the exploitation of a vulnerability, now the most common way in. (Verizon 2026 DBIR)
  • 28.96% of vulnerabilities newly exploited in 2025 were exploited on or before the day their CVE was published. (VulnCheck State of Exploitation 2026)
  • -7 days: the estimated mean time to exploit in 2025, meaning exploitation came before a patch on average. It was 63 days in 2018. (Mandiant M-Trends 2026)
  • 43 days was the median time to fully fix a KEV vulnerability, and only 26% were fully remediated in 2025. (Verizon 2026 DBIR)

1. How many CVEs are published each year

Year CVE records published Change
2022 25,059
2023 28,961 +16%
2024 40,077 +38%
2025 48,244 +20%
2026 (Q1 to Q3) 72,812 Already 51% above all of 2025

Source: CVE.org Metrics, Published CVE Records, quarterly figures summed by Invadel.

  • 2025 averaged about 132 new CVE records a day. In 2026 the quarterly count went from 15,154 (Q1) to 20,696 (Q2) to 36,962 (Q3), about 400 a day in the third quarter. (CVE.org Metrics, 2026)
  • CNAs reserved 105,782 CVE IDs in 2026 so far, against 70,729 in all of 2025 and 52,316 in 2024. (CVE.org Metrics, Reserved CVE IDs)
  • The NVD lists 97,733 non-rejected CVEs published between January 1 and November 15, 2026, about twice the 48,150 it lists for 2025. (Invadel count of NVD JSON 2.0 feeds, feed dated November 15, 2026)
  • October 2026 was the busiest month on record in the NVD data: 17,051 CVEs, more than three times December 2025 (5,491), the busiest month of 2025. (Invadel count of NVD feeds)
  • NIST says CVE submissions rose 263% between 2020 and 2025, and that submissions in the first three months of 2026 were nearly one-third higher than a year earlier. (NIST, April 2026)
  • A few issuers produce much of the volume. In 2025 Patchstack issued 7,008 CVEs and the Linux kernel CNA 5,675; the three WordPress-focused CNAs (Patchstack, Wordfence, WPScan) issued 23% of all 2025 CVEs. In 2026 GitHub (13,051) and VulnCheck (10,585) lead. (Invadel count of NVD feeds)

How we counted: we downloaded all 25 NVD yearly feeds (425,782 records), grouped them by NVD published date and excluded Rejected records. The NVD and CVE.org totals agree within 0.3% for 2025 and for each quarter of 2026.

What it means: no team reviews 400 CVEs a day by hand. What matters is which ones touch software you run and which are being exploited, the case for risk-based vulnerability management.

2. Severity distribution and the most common weaknesses

  • Of the 44,521 CVEs published in 2025 that carry a CVSS v3.1 score, 13.5% were Critical (6,015), 35.4% High (15,747), 48.7% Medium (21,669) and 2.4% Low (1,089). (Invadel count of NVD feeds; NVD’s own score where it has one, otherwise the CNA’s score)
  • In 2026 so far the mix has shifted upward: of 87,528 scored CVEs, 12.1% are Critical and 43.5% High, so more than half (55.6%) are rated High or Critical. In 2024 the share was 48.3%. (Invadel count of NVD feeds)
  • Cross-site scripting (CWE-79) was the most common weakness in 2025 CVEs, at 8,295 records (17% of the year), followed by SQL injection (CWE-89) at 4,001. (Invadel count of NVD feeds)
  • In 2026, missing authorization (CWE-862) moved to second place with 5,427 CVEs, behind cross-site scripting (8,495). (Invadel count of NVD feeds)
  • MITRE’s 2025 CWE Top 25, built from 39,080 CVE records published June 2024 to June 2025, ranks cross-site scripting first, then SQL injection, cross-site request forgery, missing authorization and out-of-bounds write. (MITRE CWE Top 25, 2025; methodology)

What it means: severity is not exploitation. In 2025, 29 of the 245 vulnerabilities CISA added to the KEV catalog carried a Medium or Low CVSS v3.1 score (Invadel count, joining the KEV catalog to NVD scores). A program that only patches Critical and High scores will miss roughly one exploited vulnerability in eight.

3. The NVD backlog

  • NIST enriched nearly 42,000 CVEs in 2025, 45% more than in any earlier year, and said that was still not enough to keep up. Since April 15, 2026 it enriches KEV-listed CVEs first (target: one business day), then federal and “critical” software; backlogged CVEs published before March 1, 2026 were moved to “Not Scheduled”. (NIST, April 2026)
  • Of the 97,733 CVEs the NVD published in 2026 through November 15, 37.2% have the status Analyzed, 38.2% Deferred, and 18.6% (18,189) are still Received, Awaiting Analysis or Undergoing Analysis. (Invadel count of NVD feeds)
  • Only 14.9% of 2026 CVEs carry a CVSS v3.1 score issued by NVD itself, down from 34.8% of 2025 CVEs. NIST no longer routinely adds its own score when the CNA has supplied one. (Invadel count of NVD feeds; policy from NIST, April 2026)
  • 43.4% of 2026 CVEs have CPE applicability data (the machine-readable list of affected products that scanners match on), against 62.7% of 2025 CVEs. (Invadel count of NVD feeds)

What it means: if your scanner relies on NVD product matching, a growing share of new CVEs will not be flagged against your assets on day one. Follow vendor advisories for internet-facing products and treat a clean scan as a partial answer; see penetration testing vs vulnerability scanning.

See what a real pentest finds. Read our sample report, then get a fixed quote in writing.

4. The CISA KEV catalog: size and additions per year

All figures in this section are Invadel counts of the CISA KEV catalog JSON, version 2026.11.13.

  • The catalog holds 1,781 vulnerabilities. Additions by year: 311 (2021, launch year), 555 (2022), 187 (2023), 186 (2024), 245 (2025) and 297 (2026, through November 13), so 2025 was up 32% on 2024. (CISA KEV catalog)
  • 361 entries (20%) are flagged as known to be used in ransomware campaigns, including 31 of the 2025 additions and 27 of the 2026 additions. (CISA KEV catalog)
  • Microsoft has the most entries (391, 22% of the catalog), followed by Cisco (103), Apple (96), Adobe (83) and Google (77). (CISA KEV catalog)
  • Only 0.41% of the CVEs published in 2025 (197 of 48,150) are in the KEV catalog as of November 2026. (Invadel count, KEV joined to NVD feeds)
  • Deadlines got shorter. Since BOD 26-04 replaced BOD 22-01 on June 10, 2026, 128 of the 164 new entries (78%) carry a 3-day federal due date and the rest 14 days. Between January 2025 and June 9, 2026, the norm was 21 days (270 of 378 entries). (CISA KEV catalog; BOD 26-04)
  • VulnCheck’s own exploitation list recorded 884 newly exploited vulnerabilities across 518 vendors in 2025, against CISA’s 245 across 99 vendors. (VulnCheck State of Exploitation 2026)

What it means: the KEV list is small enough to act on. For the current month’s entries ranked by urgency, see our October 2026 exploited vulnerabilities roundup.

5. Time to exploit and zero-days

  • Google estimates the mean time to exploit fell from 63 days in 2018 to -1 day in 2024 and -7 days in 2025. A negative number means exploitation, on average, came before a patch was released. (Mandiant M-Trends 2026)
  • 28.96% of the vulnerabilities VulnCheck saw exploited for the first time in 2025 were exploited on or before their CVE publication day, up from 23.6% in 2024. In the first half of 2026 the share was 23.43% of 495. (VulnCheck 2026; VulnCheck 1H 2026)
  • The median time from CVE publication to VulnCheck KEV listing fell from 120 days in 2025 to 80 days in the first half of 2026. (VulnCheck 1H 2026)
  • For vulnerabilities CISA added in 2026, the median gap between NVD publication and KEV listing is 12 days, down from 26 days for 2025 additions. 20% were listed on or before their NVD publication date, 60% within 30 days, and 17% more than a year later. (Invadel count, CISA KEV dateAdded against NVD published)
  • Google tracked 90 zero-days exploited in the wild in 2025, up from 78 in 2024 and below the 2023 record of 100. (Google Threat Intelligence Group, 2025 Zero-Days in Review)
  • 43 of those zero-days (48%) targeted enterprise software and appliances, up from 36 (46%) in 2024, and operating systems accounted for 39 (44%). (GTIG 2025)
  • For the first time, Google attributed more zero-day exploitation to commercial surveillance vendors than to traditional state espionage groups; 9 zero-days were used by financially motivated groups. (GTIG 2025)
  • In WordPress, about half of high-impact vulnerabilities were exploited within 24 hours, and the activity-weighted median time to first exploit was 5 hours. (Patchstack, State of WordPress Security in 2026)
  • The KEV-to-CVE ratio is falling: VulnCheck’s newly exploited vulnerabilities equaled 2.7% of new CVEs in the second half of 2023 and 1.4% in the first half of 2026, as CVE volume grew 45% in six months against 10% growth in exploited ones. (VulnCheck 1H 2026)

What it means: for internet-facing systems the patch window is days, sometimes hours, so a monthly cycle is too slow for the perimeter. Our vulnerability remediation guide covers triage by KEV and exposure rather than CVSS alone.

6. Exploitation as an initial access vector

  • Exploitation of vulnerabilities was the initial access vector in 31% of breaches in Verizon’s 2026 dataset (19,905 breaches with a known vector, drawn from more than 22,000 breaches in 145 countries), ahead of credential abuse at 13%. It was 20% in the 2025 report. (Verizon 2026 DBIR; Verizon 2025 DBIR)
  • By industry, exploitation was the initial access vector in 42% of retail breaches, 40% in public administration, 38% in manufacturing, 34% in education, 22% in finance and 20% in healthcare. (Verizon 2026 DBIR)
  • Exploits were the most common initial infection vector in Mandiant’s 2025 investigations for the sixth year running, at 32% of cases with an identified vector, against 11% for voice phishing and 9% for stolen credentials. (Mandiant M-Trends 2026, based on more than 500,000 hours of 2025 investigations)
  • The most frequently exploited vulnerabilities in Mandiant’s 2025 cases were zero-days in internet-facing web application servers: SAP NetWeaver (CVE-2025-31324), Oracle E-Business Suite (CVE-2025-61882) and Microsoft SharePoint (CVE-2025-53770). (Mandiant M-Trends 2026)
  • ENISA counted vulnerability exploitation as the intrusion vector in 21.3% of 4,875 EU incidents between July 2024 and June 2025, second to phishing at 60%. (ENISA Threat Landscape 2025)
  • Not every dataset agrees: in Sophos’ survey of 2,158 IT and security leaders in 17 countries whose organizations were hit by ransomware, exploited vulnerabilities fell 14 points to 18% of root causes, the first year in four that they were not the top cause. (Sophos State of Ransomware 2026)

What it means: Verizon and Mandiant both put exploitation first. An external network penetration test that confirms which exposed services are actually exploitable is the most direct check against that entry point.

7. Patch and remediation times

  • Only 26% of vulnerabilities in the CISA KEV catalog were fully remediated by organizations in 2025, down from 38% the year before. (Verizon 2026 DBIR)
  • The median time to full resolution of a KEV vulnerability rose to 43 days, from 32 days. In the median case, organizations had 50% more critical vulnerabilities to patch than a year earlier. (Verizon 2026 DBIR)
  • Across all vulnerabilities found in scans, only 9% were fully remediated over the year in Verizon’s 2025 data, which covered 10,000 companies. (Verizon 2025 DBIR)
  • High and critical application and API vulnerabilities took an average of 54.81 days to fix in 2025, against 39 days for device and network vulnerabilities. (Edgescan Vulnerability Statistics Report 2026, data from more than 250 companies in 30 industries)
  • 82% of organizations carry security debt, an 11% increase in one year. (Veracode State of Software Security 2026)
  • 46% of WordPress plugin and theme vulnerabilities had no fix from the developer by the time they were publicly disclosed. (Patchstack 2026)

What it means: a 43-day median fix time against a 12-day median from publication to KEV listing is the gap attackers use. Measure it in your own environment with vulnerability management metrics such as mean time to remediate and KEV closure rate.

8. Edge devices and VPNs

  • Edge devices and VPNs were the target in 22% of vulnerability exploitation actions in Verizon’s 2025 report, up almost eight-fold from 3% the year before. (Verizon 2025 DBIR)
  • Only about 54% of the edge device vulnerabilities Verizon sampled were fully remediated during the year, with a median of 32 days to do so. (Verizon 2025 DBIR)
  • The median time from CVE publication to KEV listing was 5 days across 292 KEV vulnerabilities and zero for 17 sampled edge device flaws. (Verizon 2025 DBIR)
  • Security and networking products accounted for 21 of the 43 enterprise zero-days in 2025, and at least 14 zero-days affected edge devices, a figure Google says likely understates the activity. (GTIG 2025)
  • Network edge devices (firewalls, VPNs and proxies) were the most frequently targeted technology among vulnerabilities newly exploited in 2025. (VulnCheck 2026)
  • Among ransomware attacks that began with an exploited vulnerability, stolen credentials or brute force, 21% started at a firewall and 8% at a VPN, against 38% at other exposed applications and systems. (Sophos 2026, 2,158 respondents)
  • Network and security vendors are a steady presence in KEV: Cisco (21), Fortinet (9), Ivanti (8), Citrix (6) and Check Point (5) entries were added in 2026 through November 13. (Invadel count of the CISA KEV catalog)

What it means: edge devices cannot run endpoint detection, face the internet by design and are often patched last. Put them at the top of the remediation queue and in scope for every external test; our list of common network vulnerabilities covers what we find on them.

9. Web application and open source vulnerability prevalence

  • OWASP’s Top 10:2025 is built from data on more than 2.8 million applications and 589 CWEs. Eight of the ten categories come from that data and two from a community survey. (OWASP Top 10:2025, Introduction)
  • 100% of the applications tested had some form of broken access control, the number one category, with 1,839,701 recorded occurrences. All of them also had some form of security misconfiguration (A02, up from fifth place). (OWASP A01:2025; OWASP A02:2025)
  • Injection (A05) maps to 62,445 CVEs, the most of any category, and cross-site scripting alone accounts for more than 30,000 of them. (OWASP A05:2025)
  • More than 20% of the internet-facing vulnerabilities Edgescan found in 2025 were critical or high severity. (Edgescan 2026)
  • 11,334 new WordPress ecosystem vulnerabilities were reported in 2025, up 42% on 2024. 91% were in plugins, 9% in themes, and 6 in WordPress core. (Patchstack 2026)
  • 87% of the 947 commercial codebases Black Duck audited (November 2024 to October 2025) contained at least one open source vulnerability, 44% contained a critical-risk one, and the mean codebase had 581 vulnerabilities, up from 280. 92% contained components four or more years out of date. (Black Duck OSSRA 2026)

What it means: access control and misconfiguration flaws are logic problems that scanners rarely confirm, which makes them the core of a manual web application penetration test.

What to expect in 2027

These are trends in the data above, not forecasts by the sources unless stated.

  • More CVEs. Reserved CVE IDs (105,782 in 2026 so far) and the Q3 2026 publication rate both point to another record year. NIST said in April 2026 it does not expect the growth trend “to let up anytime soon”. (NIST, April 2026; CVE.org Metrics)
  • Thinner NVD enrichment. Under NIST’s new priorities, the share of new CVEs with NVD scores and product data fell sharply in 2026, so scanners and teams will lean more on CNA data and vendor advisories. (NIST, April 2026)
  • Exploited vulnerabilities grow far more slowly than CVEs. VulnCheck’s KEV-to-CVE ratio roughly halved between 2023 and 2026, which makes exploitation evidence more useful as a filter. (VulnCheck 1H 2026)
  • Shorter federal deadlines. Most KEV entries added since June 2026 carry a 3-day federal due date, against 21 days before. (Invadel count of the CISA KEV catalog)

Part of our cybersecurity statistics hub. See also penetration testing statistics and cloud security statistics.

Sources

Frequently asked questions

How many CVEs were published in 2025 and 2026? 48,244 CVE records were published in 2025, up 20% from 40,077 in 2024, about 132 a day. In 2026, 72,812 were published in the first three quarters, 36,962 of them in the third quarter alone (CVE.org).

What percentage of vulnerabilities are actually exploited? A small fraction. Only 0.41% of CVEs published in 2025 (197 of 48,150) were in the CISA KEV catalog by November 2026. VulnCheck’s broader list puts newly exploited vulnerabilities at 1.4% of new CVEs in the first half of 2026.

How fast are vulnerabilities exploited after disclosure? Often on or before the day of disclosure: 28.96% of vulnerabilities first exploited in 2025 (VulnCheck). Google estimates the mean time to exploit in 2025 at -7 days, so on average exploitation came before a patch.

How many vulnerabilities are in the CISA KEV catalog? 1,781 as of catalog version 2026.11.13. CISA added 245 in 2025 and 297 in 2026 through November 13.

How long do organizations take to patch exploited vulnerabilities? A median of 43 days to fully remediate a KEV vulnerability in Verizon’s 2026 DBIR, up from 32, and only 26% were fully remediated in 2025.

Mark Kiss

Written by

Mark Kiss, Founder & CEO

15+ years in the field. Focus: Offensive security & strategy. Part of the team that scopes, tests, and reports for our clients. Full profile →

Find out what an attacker sees.

Tell us what to test and see your fixed price.

Prefer the full scoping questionnaire? 
Start the conversation

Get a quick proposal

We are online