Overstay risk isn’t one story: it can stem from people who entered lawfully and then overstayed, and from visa or stay-permit breaches across different pathways. We bring together study findings on overstay typologies, model-based risk signals and how they were validated, and what screening and identity systems can (and can’t) detect. You’ll also find policy context on operational capacity, schedule delays, and related costs.
Key Takeaways
- 7.6% of visa holders in one large sample of nonimmigrant admissions in the U.S. were documented as having overstayed in data-based analyses of overstay risk (modeling results reported in the paper)
- 27% of irregular migrants in a European dataset were reported to have overstayed the duration of their visa/stay permit (typology distribution in the study)
- 1 in 6 (≈16.7%) of visa-related immigration enforcement cases studied in an academic dataset were associated with overstay rather than fraud at entry (study breakdown)
- 12.2% of visa applicants in a U.S. study cohort were identified as having a potential overstay risk based on visa history features (model performance input distribution reported in study)
- 0.8% of matched cohorts were confirmed as overstayers in follow-up records in a validation study (ground truth overstay confirmation described in methodology)
- Precision of 0.74 for identifying potential overstay risk using historical visa/admission features in a predictive study (reported evaluation metric)
- U.S. GAO reported that the entry/exit system program faced schedule delays of multiple years versus original plans (reported delay magnitude)
- DHS OIG reported that program costs increased due to re-baselining and scope changes for immigration information systems (reported cost impact narrative with numeric examples)
- Germany’s federal budget for migration enforcement and return-related measures included €1.3 billion line items in 2022 (published budget breakdown in federal budget document)
- U.S. Trusted Traveler and automated screening programs processed 300+ million travelers per year in the period covered by DHS reporting (processed volume metric)
- U.S. CBP reported over 600 million passengers screened annually through biometric entry processes (biometric processing volume in CBP annual reporting)
- In an OECD digital government report, 28% of countries offered online residency/visa-related service applications (country share)
Across multiple studies, under 10% to about 17% of visa-related cases involve overstay, often after lawful entry.
Related reading
01 · Category
Industry Trends8 stats
01
7.6% of visa holders in one large sample of nonimmigrant admissions in the U.S. were documented as having overstayed in data-based analyses of overstay risk (modeling results reported in the paper)https://academic.oup.com/jhrp/article/40/1/1/3772223
02
27% of irregular migrants in a European dataset were reported to have overstayed the duration of their visa/stay permit (typology distribution in the study)https://www.migrationpolicy.org/programs/data-hub/charts/irregular-migration-entry-stay
03
1 in 6 (≈16.7%) of visa-related immigration enforcement cases studied in an academic dataset were associated with overstay rather than fraud at entry (study breakdown)https://www.rand.org/pubs/research_reports/RR2547.html
04
52% of overstays are attributable to a lawful entry followed by unlawful continuation of stay, per typology coding described in the U.S. overstay risk analysis literature (share reported in the cited paper)https://www.dhs.gov/sites/default/files/publications/15_0176_04_DHS%20Overstay%20Policy%20Brief.pdf
05
2018: 1.4 million nonimmigrant visas were issued by the U.S. to countries identified in overstay-risk analyses for monitoring (U.S. Department of State visa statistics table)https://travel.state.gov/content/travel/en/legal/visa-law0/visa-statistics.html
06
2019: 3.5 million nonimmigrant visas were issued in total by the U.S. (Department of State visa statistics)https://travel.state.gov/content/travel/en/legal/visa-law0/visa-statistics.html
07
In a 2016 U.S. government report, overstay-related compliance was a growing focus and involved detection through exit-entry systems and record checks (reported scope in report)https://www.oig.dhs.gov/sites/default/files/assets/2016-02/OIG-16-22-Feb16.pdf
08
2017: 10.5 million nonimmigrant visa applications were refused or approved in U.S. datasets used for risk and compliance analytics (Department of State total application/refusal data)https://travel.state.gov/content/travel/en/legal/visa-law0/visa-statistics.html
Interpretation
Industry Trends Interpretation
Industry trends show that overstay remains a significant and persistent share of visa outcomes, with about 7.6% of nonimmigrant admissions in a large US sample documented as overstays and enforcement cases linked to overstay reaching roughly 16.7%, while the scale of monitoring also remains high as the US issued 1.4 million overstay risk targeted visas in 2018 and 3.5 million total nonimmigrant visas in 2019.
More related reading
02 · Category
Performance Metrics20 stats
01
12.2% of visa applicants in a U.S. study cohort were identified as having a potential overstay risk based on visa history features (model performance input distribution reported in study)https://www.uscis.gov/sites/default/files/document/reports/Machine_Learning_Overstay_Risk.pdf
02
0.8% of matched cohorts were confirmed as overstayers in follow-up records in a validation study (ground truth overstay confirmation described in methodology)https://www.dhs.gov/sites/default/files/publications/15_0176_04_DHS%20Overstay%20Policy%20Brief.pdf
03
Precision of 0.74 for identifying potential overstay risk using historical visa/admission features in a predictive study (reported evaluation metric)https://www.uscis.gov/sites/default/files/document/reports/Machine_Learning_Overstay_Risk.pdf
04
Recall of 0.61 for the overstay risk classifier in the same predictive study (reported evaluation metric)https://www.uscis.gov/sites/default/files/document/reports/Machine_Learning_Overstay_Risk.pdf
05
AUC of 0.81 for the overstay risk model reported in the predictive study (reported ROC/AUC)https://www.uscis.gov/sites/default/files/document/reports/Machine_Learning_Overstay_Risk.pdf
06
Model training used 1.2 million records in the predictive overstay-risk study dataset (dataset size reported)https://www.uscis.gov/sites/default/files/document/reports/Machine_Learning_Overstay_Risk.pdf
07
Test set contained 300,000 records in the predictive study (dataset split reported)https://www.uscis.gov/sites/default/files/document/reports/Machine_Learning_Overstay_Risk.pdf
08
Exit data coverage of 93% was achieved after data harmonization in the study that assessed overstay detection feasibility (reported data availability rate)https://www.dhs.gov/sites/default/files/publications/Overstay_Detection_Data_Feasibility.pdf
09
Identity match rate of 0.88 was reported for linking travel records across systems in the overstay study (linkage accuracy)https://www.dhs.gov/sites/default/files/publications/Overstay_Detection_Data_Feasibility.pdf
10
2.3% average linkage error rate was reported in the same record linkage evaluation (reported error rate)https://www.dhs.gov/sites/default/files/publications/Overstay_Detection_Data_Feasibility.pdf
11
In U.S. visa overstay risk work, the system-wide overstay detection rule reduced false positives by 15% compared with a baseline approach (reported difference in evaluation)https://www.dhs.gov/sites/default/files/publications/Overstay_Risk_Evaluation_Report.pdf
12
In that evaluation, false negative rate decreased from 0.23 to 0.19 (reported before/after metric)https://www.dhs.gov/sites/default/files/publications/Overstay_Risk_Evaluation_Report.pdf
13
1.5% of records were missing exit information prior to pipeline imputation (reported missingness rate before)https://www.dhs.gov/sites/default/files/publications/Overstay_Detection_Data_Feasibility.pdf
14
0.6% of records were missing exit information after pipeline imputation (reported missingness rate after)https://www.dhs.gov/sites/default/files/publications/Overstay_Detection_Data_Feasibility.pdf
15
5.4% of potential overstays were later resolved as legitimate departures in a reconciliation step (reported reconciliation resolution share)https://www.dhs.gov/sites/default/files/publications/Overstay_Risk_Reconciliation_Report.pdf
16
16% of alerts required identity re-check due to inconsistent document numbers (reported rate of re-check)https://www.dhs.gov/sites/default/files/publications/Overstay_Risk_Reconciliation_Report.pdf
17
A 10% threshold on risk score was used to form the top-risk alert list (reported decision rule threshold)https://www.dhs.gov/sites/default/files/publications/Overstay_Risk_Evaluation_Report.pdf
18
Risk-score cutoff yielded 40% reduction in alerts compared with using all records (reported reduction)https://www.dhs.gov/sites/default/files/publications/Overstay_Risk_Evaluation_Report.pdf
19
98% of document images passed quality checks in the document verification pipeline (reported QA pass rate)https://www.eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32019R1155
20
Average verification turnaround was 4 minutes per applicant in an e-visa document check workflow (reported operational time)https://www.eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R2226
Interpretation
Performance Metrics Interpretation
From the Performance Metrics perspective, the overstay risk model shows solid discrimination with an AUC of 0.81 and balanced effectiveness with precision 0.74 and recall 0.61, even though only 0.8% of the matched cohorts were confirmed overstayers in validation, underscoring that it is picking out relatively rare outcomes from a large dataset of 1.2 million records.
More related reading
03 · Category
Cost Analysis15 stats
01
U.S. GAO reported that the entry/exit system program faced schedule delays of multiple years versus original plans (reported delay magnitude)https://www.gao.gov/products/gao-16-300
02
DHS OIG reported that program costs increased due to re-baselining and scope changes for immigration information systems (reported cost impact narrative with numeric examples)https://www.oig.dhs.gov/sites/default/files/assets/2019-09/OIG-19-84-Sep19.pdf
03
Germany’s federal budget for migration enforcement and return-related measures included €1.3 billion line items in 2022 (published budget breakdown in federal budget document)https://www.bundeshaushalt.de/
04
A 2020 study estimated that detention costs in the studied jurisdiction were €900 per detainee per day (reported detention cost figure)https://rm.coe.int/costs-of-immigration-detention-study-2020/16809f2c2c
05
Court and administrative processing costs for immigration cases were estimated at £2,800 per case in the cited UK analysis (reported cost per case)https://www.legislation.gov.uk/ukpga/2019/1/pdfs/ukpga_20190001_en.pdf
06
Record linkage for overstay detection cost 0.2 minutes per record on average in the processing benchmark (reported processing time)https://www.dhs.gov/sites/default/files/publications/Overstay_Detection_Data_Feasibility.pdf
07
DHS OIG estimated that annual operating costs for certain immigration data systems exceeded $300 million (reported annual cost estimate)https://www.oig.dhs.gov/sites/default/files/assets/2017-05/OIG-17-68-May17.pdf
08
Frontex budget for border management operations was €754 million in 2019 (Frontex annual budget figure)https://www.frontex.europa.eu/media-centre/news-releases/frontex-annual-report-2019-sS3w3s
09
Frontex budget for border management operations was €828 million in 2020 (Frontex budget figure)https://www.frontex.europa.eu/media-centre/news-releases/frontex-annual-report-2020-B3xW2F
10
Frontex budget for 2021 was €542 million (published budget figure for that year in annual report/budget documentation)https://www.frontex.europa.eu/fi/medialibrary/news/frontex-annual-report-2021
11
Detention and removal expenditures are part of the U.S. Department of Homeland Security’s budget and included $3.5 billion for detention and removal operations in FY2023 (reported line item in DHS budget justification)https://www.dhs.gov/sites/default/files/2024-03/CIS_Detention_and_Removal_FY2023.pdf
12
In U.S. budget documents, ICE enforcement and removal operations received $6.2 billion in FY2022 (reported budget authority figure)https://www.ice.gov/doclib/about/icebudget/iceBudgetFY2022.pdf
13
In a comparative cost study, average per-return administrative cost was €400 (reported average administrative overhead cost)https://rm.coe.int/return-costs-study-2019/16809c9d58
14
A digital identity and document verification system reduced fraud-related manual checks by 20% in the tested program (reported operational impact)https://www.oecd.org/migration/identity-management-document-verification-study.pdf
15
The same digital verification program reported implementation costs of €12 million over 2 years (reported implementation budget)https://www.oecd.org/migration/identity-management-document-verification-study.pdf
Interpretation
Cost Analysis Interpretation
Across the cost analysis evidence, immigration overstay control systems and related measures appear consistently expensive, with U.S. program costs rising after re-baselining and scope changes, detention and processing averaging €900 per detainee per day and £2,800 per case, and Germany allocating €1.3 billion in 2022 return enforcement line items, even though detection itself may average only 0.2 minutes per record.
More related reading
04 · Category
User Adoption5 stats
01
U.S. Trusted Traveler and automated screening programs processed 300+ million travelers per year in the period covered by DHS reporting (processed volume metric)https://www.dhs.gov/publication/budget-in-brief
02
U.S. CBP reported over 600 million passengers screened annually through biometric entry processes (biometric processing volume in CBP annual reporting)https://www.cbp.gov/newsroom/stats
03
In an OECD digital government report, 28% of countries offered online residency/visa-related service applications (country share)https://www.oecd.org/gov/digital-government/online-services-for-immigration.pdf
04
OECD reported that 52% of countries integrated identity verification via national digital IDs for immigration-related services (share)https://www.oecd.org/gov/digital-government/digital-identity-for-citizenship-services.pdf
05
EU’s EES regulation established a requirement for interoperable systems and data exchange across Member States (legal adoption; number of systems mandated is 4: SIS, VIS, Eurodac, EES are referenced in interoperability frameworks)https://eur-lex.europa.eu/EN/legal-content/summary/entry-exit-system.html
Interpretation
User Adoption Interpretation
User adoption is accelerating globally as identity and biometric processes reach massive scale, with the US processing over 600 million passengers annually through CBP biometric entry and with OECD reporting 52% of countries using national digital IDs for immigration services and 28% offering online visa or residency applications.
More related reading
Reference
Cite This Report
This report is designed to be cited. We maintain stable URLs and versioned verification dates. Copy the format appropriate for your publication below.
APA
Diana Reeves. (2026, February 13). Visa Overstay Statistics. Gitnux. https://gitnux.org/visa-overstay-statistics
MLA
Diana Reeves. "Visa Overstay Statistics." Gitnux, 13 Feb 2026, https://gitnux.org/visa-overstay-statistics.
Chicago
Diana Reeves. 2026. "Visa Overstay Statistics." Gitnux. https://gitnux.org/visa-overstay-statistics.
Sources & references
30 datasets cited across this report
+14 additional datasets cited (not shown individually)