Password advice is full of recycled numbers with no date or source, which makes it hard to argue for the controls that actually reduce password risk. This page collects current password and credential hygiene figures from the publishers' own reports: the Verizon 2026 Data Breach Investigations Report (DBIR), the UK government's Cyber Security Breaches Survey 2025, the UK National Cyber Security Centre (NCSC) analysis of breached passwords, Have I Been Pwned (HIBP), Microsoft's identity research and NIST SP 800-63B-4. Every figure carries its publisher and data year in the same sentence or table row. It is for identity, security awareness and policy owners who need evidence for a password policy change, a breached-password screening project or a move to passkeys. How to build the controls is covered in the Password Security Guide; this page is the evidence. Last reviewed: November 2026.
Key takeaways
- Compromised beats weak: the 2026 DBIR finds a median 4% of scanned accounts using breached passwords and says users are more than four times more likely to use an already compromised password than a "weak" one.
- Stolen credentials remain in a third of breaches: the 2026 DBIR records use of stolen credentials in 36% of breaches and credential abuse at some point in 39%.
- Attack volume keeps climbing: Microsoft observed about 7,000 password attacks per second in 2024, more than double the 2023 rate.
- Policy exists, MFA lags: 73% of UK businesses have a strong password policy but only 40% use two-factor authentication (Cyber Security Breaches Survey 2025).
- The standard has moved: NIST SP 800-63B-4 sets a 15-character minimum for single-factor passwords and forbids composition rules and periodic forced changes.
Headline statistics
| Statistic | Figure | Source, publisher and data year |
|---|---|---|
| Breaches involving use of stolen credentials | 36% | 2026 DBIR, Verizon (incidents 1 November 2024 to 31 October 2025) |
| Credential abuse as initial access vector | 13% (22% in the prior edition) | 2026 DBIR, Verizon |
| Median share of scanned accounts with breached passwords | 4% (n=7,345) | 2026 DBIR, Verizon |
| Median share of users reusing or sharing passwords | About 6% | 2026 DBIR, Verizon |
| Password attacks observed by Microsoft | About 7,000 per second | Microsoft Security Blog, May 2025 (2024 data) |
| Reduction in account compromise risk with MFA | 99.22%; 98.56% where credentials had leaked | Meyer et al., Microsoft, arXiv 2305.00945 (2023) |
| UK businesses with a strong password policy | 73% (57% of charities) | Cyber Security Breaches Survey 2025, DSIT (fieldwork August to December 2024) |
| Accounts in breach data using 123456 | 23.2 million | NCSC, April 2019 |
| Unique passwords in one credential stuffing corpus | 1.3 billion, 625 million new to HIBP | Have I Been Pwned, Troy Hunt, November 2025 |
Statistics by theme
Passwords in breaches
The DBIR shows that counting only the first step of a breach understates password risk. Verizon notes that part of the fall from 22% to 13% comes from adding pretexting as a tracked vector; without that change the figure would have been 16%.
| Measure (2026 DBIR, Verizon) | Figure |
|---|---|
| Credential abuse as initial access | 13% (16% on the previous classification) |
| Credential abuse at any point in the breach | 39%, the top vector on that count |
| Use of stolen credentials | 36% of breaches |
| Credentials among data compromised | 28% of breaches |
| Basic Web Application Attacks | 3,217 incidents, 2,281 with confirmed disclosure; credentials 52% of data compromised |
| Credentials compromised in Social Engineering breaches | 39% |
| Credentials compromised in System Intrusion breaches | 26% |
Microsoft's Digital Defense Report (2026 edition) adds what happens after the first login: 52.2% of valid account intrusions it describes involved follow-on credential theft, and 89% to 95% of email phishing attachments it analysed led to a credential theft effort.
Hygiene inside organisations
The DBIR's password hygiene figures come from scans of contributor environments and are medians across organisations, which makes them useful benchmarks for your own audit.
| Hygiene measure (2026 DBIR) | Median | What it tells you |
|---|---|---|
| Accounts with breached passwords | 4% of accounts scanned | Breached-password screening finds real exposure |
| Active Directory accounts not meeting complexity requirements | Under 1% | Complexity rules are already met; they are not the gap |
| Users reusing or sharing passwords | About 6% | Reuse is common enough to target with screening and training |
| Credential leak events per year, small organisations | 7 | Even small firms see repeated leaks |
| Credential leak events per year, larger organisations | About 20 | Leak monitoring needs a process, not ad hoc checks |
Together these figures make the central point of current password policy: accounts pass complexity rules and still use passwords that attackers already hold. Of ransomware victims that had an infostealer or credential leak event in the prior year, half had it within 95 days of the attack, which gives defenders a window to reset exposed credentials. Third-party cloud environments show slow remediation: for weak passwords and permission misconfigurations, it took almost eight months to resolve half the findings, and 37% of organisations had an IaaS admin account without MFA at the time of the snapshot.
The most common passwords
The NCSC's 2019 analysis of passwords in public breach data remains the most cited primary source on password choice. Its counts are accounts in breach corpora, not current users.
| Category (NCSC, 2019) | Most used | Accounts in breach data |
|---|---|---|
| Overall | 123456 | 23.2 million |
| Overall, second to fifth | 123456789, qwerty, password, 1111111 | 7.7m, 3.8m, 3.6m, 3.1m |
| Names | ashley | 432,276 |
| Football teams | liverpool | 280,723 |
| Musicians | blink182 | 285,706 |
| Fictional characters | superman | 333,139 |
The same NCSC release reported that fewer than half of respondents always use a strong, separate password for their main email account, the account that resets every other account.
Credential supply
- Corpus size: the credential stuffing data set loaded into HIBP on 5 November 2025 held 1,957,476,021 unique email addresses and 1.3 billion unique passwords, 625 million of them never seen by the service before. One subscriber confirmed that a password in the data was one they had used 10 to 20 years earlier, a reminder that old passwords stay in circulation.
- Screening demand: HIBP's Pwned Passwords service handles more than 18 billion requests a month, a measure of how widely breached-password checks are already built into sign-up and change flows. See Breached password checks.
- Reuse in practice: the 23andMe credential stuffing breach began with passwords reused from other sites: more than 18,000 accounts were accessed directly, data on almost 7 million people was exposed through a relatives feature, and fewer than 22% of customers had MFA or single sign-on turned on.
Policy and MFA adoption
| Control (Cyber Security Breaches Survey 2025, UK) | Businesses | Charities |
|---|---|---|
| Policy ensuring users set strong passwords | 73% | 57% |
| Two-factor authentication for networks or applications | 40% (92% of large businesses) | 35% |
| Any breach or attack in the last 12 months | 43% | 30% |
| Phishing among those with a breach or attack | 85% | 86% |
The gap between password policy and MFA is the clearest finding for smaller organisations: most have rules about password strength, fewer than half have the control Microsoft's research credits with a 99.22% reduction in compromise risk. The survey sampled 2,180 businesses, 1,081 charities and 574 education institutions. See the MFA Adoption Statistics.
Passkeys as the outcome measure
Microsoft reported in May 2025 that passkey sign-ins succeed about 98% of the time against about 32% for passwords, are eight times faster than password plus MFA, and that nearly one million passkeys were registered for Microsoft accounts each day. These are vendor figures for consumer accounts, but they are the strongest published evidence that removing the password improves both security and usability. See the Passkey Adoption Statistics.
What the standard now says
NIST SP 800-63B-4, published in August 2025 and superseding SP 800-63B, turns the evidence above into requirements. Many corporate policies still reflect the older rules, so it is worth comparing directly.
| Topic | SP 800-63B-4 requirement | Evidence on this page |
|---|---|---|
| Minimum length | 15 characters when the password is the only factor; 8 when used within MFA | Short common passwords dominate breach data (NCSC) |
| Maximum length | SHOULD permit at least 64 characters | Supports passphrases and password managers |
| Composition rules | SHALL NOT require mixtures of character types | Under 1% of AD accounts fail complexity yet 4% are breached (DBIR) |
| Periodic change | SHALL NOT require periodic changes; SHALL force a change on evidence of compromise | Half of credential events precede ransomware by 95 days or less (DBIR) |
| Blocklist | Compare new passwords against common, expected or compromised values | 1.3 billion unique passwords in one corpus (HIBP) |
| Hints and security questions | Not permitted | Recovery paths are account takeover paths |
See the Password Policy Guide for policy wording.
What the numbers mean for defenders
| Finding | Implication | Control and framework reference |
|---|---|---|
| 4% breached, under 1% non-complex (DBIR) | Screen against breach corpora rather than tighten complexity | NIST SP 800-53 IA-5(1); NIST CSF 2.0 PR.AA-01 |
| 7,000 password attacks per second (Microsoft) | Throttle and detect spraying and stuffing at every sign-in endpoint | SP 800-53 AC-7; ATT&CK T1110.003, T1110.004 |
| Stolen credentials in 36% of breaches (DBIR) | Treat any password-only account as exposed | ATT&CK T1078; SP 800-53 IA-2 |
| Follow-on credential theft in 52.2% of intrusions (Microsoft) | Watch for credential access after the first login | ATT&CK T1555.003; SP 800-53 SI-4 |
| 2FA at 40% of UK businesses (CSBS) | MFA is the largest remaining gain for small organisations | NIST CSF 2.0 PR.AA-03 |
For a small organisation the order is simple: turn on MFA everywhere it is offered, add a password manager and a breached-password check, and drop forced expiry. Large organisations should benchmark their own breached-password and reuse rates against the DBIR medians, then move privileged and high-value users to passkeys first. Passwords held by service accounts and scripts are not covered by these surveys but follow the same logic; see the Service Account Password Rotation Runbook.
Metrics to track internally
- Breached-password rate: share of active accounts matching a known breached password, against the DBIR's 4% median.
- Reuse rate: share of accounts sharing a password hash with another account, against the DBIR's 6% median.
- MFA coverage: share of human accounts with MFA enforced, split by phishing-resistant and other methods.
- Passwordless share: share of sign-ins completed with a passkey or security key.
- Leak-to-reset time: days from a credential appearing in a leak or infostealer feed to the reset.
- Password-related help desk tickets: resets per 100 users per month, which should fall as passkeys spread.
Common failure patterns when using password statistics
- Quoting 2019 NCSC counts as current behaviour: they count accounts in breach data collected over years, not what people choose today.
- Reading the DBIR's fall to 13% as falling risk: part is a classification change, and stolen credentials still appear in 36% of breaches.
- Using corpus size as victim count: HIBP counts unique addresses and passwords in a data set, not compromised accounts.
- Citing the 99.22% MFA figure for every MFA type: the Microsoft study measured Azure Active Directory accounts and found authenticator apps outperformed SMS, so the headline figure blends methods of different strength.
- Treating a password policy as a control outcome: 73% of UK businesses have one; the DBIR shows breached passwords pass policy.
- Citing roundup pages: secondary sites often drop the year and sample. Cite the publisher, as this page does.
Choosing the right source
| Question you need to answer | Best source on this page | Why |
|---|---|---|
| How often do stolen passwords feature in breaches? | Verizon DBIR | Large multi-contributor breach data set with stated periods and sample sizes |
| How does our password hygiene compare with peers? | Verizon DBIR hygiene medians | Breached, reused and non-complex account medians from real scans |
| How many UK organisations have password and MFA controls? | Cyber Security Breaches Survey | Official statistics with business and charity splits |
| Which passwords do people actually choose? | NCSC breach analysis | Counts per password from public breach data |
| How much does MFA reduce compromise? | Meyer et al., Microsoft | Measured on a large population of real accounts |
| What should our policy require? | NIST SP 800-63B-4 | The current standard, with SHALL and SHOULD statements |
For a board paper, one breach prevalence figure (36% of breaches involve stolen credentials) and one internal benchmark (your breached-password rate against the 4% median) usually make the case better than a long list of numbers.
Methodology notes and caveats
- Verizon DBIR: incidents contributed by Verizon and partners for 1 November 2024 to 31 October 2025; hygiene figures are medians across scanned organisations.
- Cyber Security Breaches Survey 2025: UK government survey, fieldwork August to December 2024; self-reported controls.
- NCSC: analysis of passwords in public breach data, April 2019.
- Microsoft: telemetry from Microsoft's own services and customers, attributed as such.
- HIBP: counts of unique values in loaded data sets.
Practitioner checklist
- Cite each figure with publisher, report name and data year in the same sentence.
- Measure your breached-password and reuse rates and compare them with the DBIR medians.
- Screen new and changed passwords against a breached-password list.
- Align the policy to SP 800-63B-4: length, no composition rules, no periodic expiry.
- Force resets when credentials appear in leak or infostealer data.
- Enforce MFA, phishing-resistant for administrators and finance roles.
- Plan passkey rollout and track the passwordless share of sign-ins.
- Recheck figures each year when the DBIR and the Breaches Survey publish.
Standards and references
- NIST: SP 800-63B-4 Digital Identity Guidelines, Authentication and Authenticator Management
- Verizon: 2026 Data Breach Investigations Report
- DSIT: Cyber Security Breaches Survey 2025
- NCSC: Most hacked passwords revealed
- arXiv: How effective is multifactor authentication at deterring cyberattacks?
- Microsoft: Pushing passkeys forward
- Have I Been Pwned: 2 Billion Email Addresses Were Exposed
- Have I Been Pwned: Pwned Passwords
- Microsoft: Digital Defense Report
Related NHI Mgmt Group resources: Password Security Guide · Identity Attack Statistics Guide · Account Takeover Statistics · Password Spraying and Credential Stuffing Defence Guide