Latest Statistics on International Child Abduction

1. The latest annual figures: applications received in 2025

The Permanent Bureau’s Report of 2026 records incoming applications from other Contracting Parties for assistance in securing a child’s return and, separately, for organising or securing access rights. The following figures are application counts. They do not identify how many children were involved, whether the requests resulted in proceedings, or what the eventual outcomes were.

The Permanent Bureau cautions that reporting remains incomplete and that some figures are provisional or estimates. Participation also differs between years; the United Kingdom’s 2025 figures, for example, are expressly marked provisional and not yet verified. The increase in the aggregate reported total cannot, by itself, establish a corresponding worldwide increase in the incidence of abduction. Blank cells likewise cannot be read as zeros.

The 2026 publication supplements the earlier 2024 collection with later responses. Consequently, a comparison should identify the version used and maintain a consistent definition of incoming return or access applications. Combining these annual totals with the detailed 2021 outcome percentages would create a result that neither source reports.

2. The detailed 2021 study: coverage and method

Professor Nigel Lowe and Victoria Stephens conducted the fifth detailed study in consultation with the HCCH Permanent Bureau, in preparation for the Eighth Special Commission meeting of 10–17 October 2023. The updated Global Report of September 2024 records responses from 77 of the 101 States party to the Convention in 2021. It follows studies of applications made in 1999, 2003, 2008 and 2015; those are data years, not a continuous annual series.

The responding authorities supplied information on 2,590 incoming applications: 2,191 for return and 399 for access. Applications commenced between 1 January and 31 December 2021 were followed for outcomes through 30 June 2023. A proceeding still unresolved at that cut-off was classified as pending. An outcome reached in 2022 or 2023 therefore remains part of the 2021 application cohort.

The study’s authors estimated that the responses captured about 95% of applications routed through Central Authorities, using additional outgoing information and assumptions about non-responding States. That estimate concerns Convention applications, not all abductions worldwide. Direct court applications unknown to Central Authorities, cases proceeding through other international instruments, domestic abductions and situations involving non-Contracting States fall outside that asserted coverage.

The Central Authorities supplied the underlying administrative records. The cover note also acknowledges voluntary contributions from the People’s Republic of China, Germany, the Philippines and the United Kingdom, ICMEC and the US Friends of the Hague Conference Foundation. Their support helped make the study possible; the evidential basis for each finding remains the reported data and the methodology, including the limits of the information available for particular variables.

3. Taking persons, caregiving and children

For the 2,065 return applications in which the taking person’s relationship to the child was known, the report records mothers in 75%, fathers in 23%, and grandparents, institutions or other relatives in the remaining 2%. The maternal proportion was 73% in the 2015 study. This is a change in the reported proportion, not proof that the absolute number of abductions by mothers increased or that gender explains the removal or retention.

Caregiving status was available for a smaller group of applications. Among mothers for whom that information was reported, 94% were primary or joint-primary carers. Those categories must be retained: joint care is not equivalent to sole care. The findings describe the recorded family circumstances and do not determine whether a particular removal was wrongful, whether consent existed or whether an exception to return is established.

The study reports an average of 1.3 children per return application, with 74% involving a single child, and a mean age of 6.7 years. These are different measures: the number of applications, the number of children per request and the ages of the children for whom information was supplied. They should not be multiplied into a new national or global child count without the underlying data and the study’s missing-data qualifications.

The access cohort was analysed separately. The report identifies 399 access applications and a higher average child age of 8.1 years, compared with 6.7 in return applications. An access request seeks the effective exercise of contact rights under Article 21; it should not automatically be counted as another removal or retention, nor should its outcomes be pooled with those of return requests.

4. Return outcomes and the limits of the reported rate

Outcome information was available for 2,106 of the 2,191 return applications. Within that reported group, the study’s overall return rate was 39%, comprising 16% voluntary agreements for return and 23% judicial orders for return. The corresponding reported rate in the 2015 study was 45%. An order directing return and the physical implementation of that order are separate events; this classification is not a verified execution rate for every child.

The denominator includes pending and other recorded categories, rather than only finally litigated cases. At the cut-off date, 233 applications were still pending, representing 11% of the group with outcome information. Other applications had been withdrawn, rejected at the administrative stage or closed with different outcomes, including agreements that the child would remain in the requested State. Those categories cannot all be labelled failed return claims.

The report records 38% of applications with outcome information as decided by a court, and a return order in 59% of that judicial category. It identifies 283 judicial refusals in the updated version. Reasons for refusal were available in 240 of those applications, with Article 13(1)(b) the most frequently recorded ground when sole and combined grounds were considered.

That finding does not measure the success rate of invoking Article 13(1)(b). The study collected reasons in applications ending in refusal; it did not identify every case in which that exception was argued unsuccessfully or established without preventing a return order. Nor does a reference to grave risk identify, without further evidence, the particular factual allegation or the prevalence of domestic violence across all Convention applications.

5. Resolution times and the pandemic context

Overall timing information was available for 1,146 return applications. The reported mean was 207 days from receipt by the Central Authority to the final outcome, compared with 164 days in the 2015 study. In the updated 2021 report, the means for the principal outcome categories were 130 days for a voluntary return, 197 for a judicial return order and 268 for a judicial refusal. The latter figures include appeals where applicable.

These are observed averages for recorded cases, not filing-to-judgment deadlines or forecasts for new proceedings. They also do not incorporate an eventual final duration for applications still pending at the cut-off. Article 11 of the Convention requires expedition and provides a right to request reasons for delay where a decision has not been reached within six weeks of commencement of the proceedings. That legal mechanism has a different starting point and function from the study’s administrative-receipt averages.

The authors discuss COVID-19 restrictions, disrupted travel and court functioning as possible influences on the 2021 findings. Their addendum expressly treats the extent of the pandemic’s effect on outcomes and timing as a matter for discussion and speculation. The study supports examining delay and its causes; it does not establish a causal explanation for every change or demonstrate that the same pattern persists in 2025 or 2026.

6. Appeals and comparisons across studies

The study records 359 return applications involving an appeal, or 42% of the 856 applications that went to court. This court-involvement group differs from the category already decided by a court used in the outcome analysis. For the 244 applications in which both the first-instance result and the final decision were known, 81% reached the same outcome on appeal as at first instance.

The study’s reported appeal proportion was higher than in 2015, but neither 42% nor 81% is a prediction for an individual appeal. The latter percentage concerns the subset with comparable outcome information and includes both confirmation of return and confirmation of refusal. An affirmation rate cannot be converted into the applicant’s probability of recovering a child.

Historical comparisons require the same discipline. The 2021 report makes certain comparisons using only States that responded in both study years and cautions about the pandemic period. Changes in participation, missing variables, unresolved proceedings and the composition of applications affect interpretation. A series of observed proportions does not, without further analysis, establish a worldwide causal trend or a ranking of legal systems.

7. What the Brazilian figures do and do not show

The Brazil row in the HCCH annual table records 99 incoming return applications and eight incoming access applications in 2025. The comparison figures for 2024 are 74 and three. These are requests received by the Brazilian Central Authority from other Contracting Parties within the survey’s categories. They are not a count of all children abducted to or from Brazil, all proceedings pending in Brazilian courts or all returns carried out during those years.

ACAF’s institutional explanation distinguishes active and passive international cooperation requests. Article 3 of Portaria MJSP No. 688/2024 defines the active category by removal from Brazil or retention abroad, and the passive category by removal to or retention in Brazil. That direction must not be confused with the English expression active cases used to describe an unresolved caseload. New requests received during a year and the stock of pending matters are also different measures.

The MJSP statistics portal separately identifies monthly raw DRCI data and annual consolidated publications. Its published 2025 indicators cover broader criminal and civil cooperation activities. Such aggregates cannot be substituted for the HCCH’s return and access series. No additional ACAF total or national return-success percentage is inferred here from an unavailable breakdown, an unlabelled chart or the absence of a figure.

8. Using the findings in legal practice and public policy

The data make procedural delay, the availability of reliable records and effective international cooperation legitimate subjects for examination. Their practical value depends on retaining the distinction between reported applications and established wrongful removals or retentions. A particular return request remains governed by the Convention’s scope, custody-rights and habitual-residence requirements, the evidence and any applicable exception; demographic or outcome percentages do not satisfy those tests.

Article 19 also separates a Convention return decision from the merits of custody. Accordingly, a return statistic cannot establish who should ultimately exercise parental responsibility or with whom a child should live after the return proceedings. Where access is the object of the request, Article 21 and its cooperation functions require their own analysis.

For institutional evaluation, the study supports better recording of the successive administrative and judicial stages, clear definitions of outcomes and sustained cooperation among Central Authorities, courts and legal practitioners. The annual collection adds more recent information on reported demand, while the earlier detailed cohort illuminates characteristics, decisions and timing. Preserving the date, universe and limitations of each source allows both to inform policy without turning historical observations into guarantees about current cases.