Germany’s 2025 Knife-Attack Statistics: What 29,243 Cases Really Show

The total conceals the main development

In 2024, the police recorded 29,014 knife attacks. The difference in 2025 was only 229 cases, a small movement in a national annual statistic. That apparent stability disappears once the total is broken down.

Of the 29,243 cases recorded in 2025, 14,787 fell within the Police Crime Statistics aggregate known as violent crime. This was a decline of 954 cases, or 6.1 per cent. Threats involving a knife rose by 9.5 per cent to 13,748 cases, equivalent to an increase of roughly 1,200.

A further 708 cases, representing 2.4% of the overall total, relate to coercion as well as resistance to and assaults on law-enforcement officers and equivalent persons. This figure was calculated from the published overall total and the two main groups reported separately. The BKA’s summary publication does not show how the 708 cases are distributed between the individual offence categories.

The slight increase in the headline figure is therefore the result of two movements that almost cancel each other out. Far fewer knife-attack cases were recorded within violent crime, while substantially more were recorded as threats. Saying that knife attacks “rose again” is mathematically correct but materially incomplete without this distinction. It would be equally incomplete to claim that the problem clearly diminished overall. The data show a shift between different forms of offending, not a uniform trend.

The comparison with violent crime as a whole is also relevant. Total violent crime fell by 2.3 per cent to 212,335 cases in 2025. Knife attacks within that group fell by 6.1 per cent, and their share of violent crime declined from 7.2 to 7.0 per cent. The 2025 data therefore do not indicate that knives became more prevalent within recorded violent crime. This component fell faster than violent crime overall.

Threats should not be dismissed as harmless. An immediate threat with a knife can create serious danger and substantial psychological harm. Statistically, however, it is different from an attack that is carried out. Combining both in one headline number obscures that distinction. The figure of 29,243 does not mean that Germany recorded 29,243 physical knife assaults or 29,243 injured victims.

The previous years do not form one continuous national series

Uniform national recording of knife attacks followed a decision by Germany’s Conference of Interior Ministers in 2018. On 1 January 2020, “knife attack” was introduced into the Police Crime Statistics as an additional phenomenon marker. Its technical introduction did not immediately produce a valid national total.

No valid federal results could be published for 2020. Selected figures became available for 2021, when 10,917 knife attacks were recorded within violent crime. Of these, 7,071 involved dangerous or grievous bodily harm and 3,060 involved robbery. The 2022 and 2023 reports concentrated on these more specific offence groups. Knife attacks involving dangerous or grievous bodily harm increased from 7,071 in 2021 to 8,160 in 2022 and 8,951 in 2023. The robbery figures increased from 3,060 to 4,195 and then to 4,893 over the same period.

The partial series show a substantial rise in knife attacks connected with dangerous or grievous bodily harm and robbery between 2021 and 2023. They do not cover the full phenomenon on a consistent basis. Threats and other offence groups were not included in the same way. Adding selected categories retrospectively to construct an apparent national total would not be methodologically sound.

Only from 1 January 2024 did every Police Crime Statistics case have to carry a value from the phenomenon catalogue. Technical restrictions were also introduced to limit the offence codes for which “knife attack” could be selected. These changes were intended to reduce missing entries and implausible classifications. Valid and sufficiently harmonised national totals are therefore available only from the 2024 reporting year.

The fully comparable overall series currently consists of two years. The difference between 2024 and 2025 can be calculated, but it is not yet a long-term trend. Claims that knife attacks have risen nationally “for years” may rely on selected offence groups or individual state statistics. They cannot yet be demonstrated through one unchanged BKA total covering all knife attacks. Conversely, the 2025 decline within violent crime is not enough to establish a lasting reversal.

Longer comparisons of threat offences face an additional legal break. Section 241 of the German Criminal Code was substantially broadened in April 2021. The reform extended criminal liability to certain threats involving less serious offences that were not previously covered in the same way. This expanded the conduct captured by the criminal law and limits comparisons with the years before the reform. It does not directly explain the change between 2024 and 2025, because the same law applied in both years.

Definition, counting rules and the failed 2025 data extension

“Knife attack” is not a separate criminal offence. It is an additional marker attached to a police-recorded case. It covers acts in which an attack with a knife is immediately threatened against a person or is carried out. Merely carrying a knife is not sufficient.

For threats, the recording rules refer to the victim’s perception. The marker may be used when at least one victim subjectively feels threatened with a knife. This can include situations in which no object is recovered or it is initially only assumed that the object was a knife. The rule is understandable for police recording and criminal investigation, but it means that not every case in the total necessarily involves an objectively confirmed knife.

The counting system introduces further limits. The Police Crime Statistics are an output statistic: a case is generally counted when the police have completed their processing and transferred it to the prosecution service. The reporting year and the year of the offence may therefore differ. An offence committed late in 2024 may enter the 2025 statistics if police processing was completed in 2025. Annual changes can consequently reflect not only changes in offending but also processing times, late submissions and the timing of case closure.

The statistics also cover only the police-known field. Offences are included when they are reported or otherwise become known to the police. Changes in reporting behaviour, police controls or classification practice can affect the recorded total even when the underlying level of offending does not move to the same extent. Incidents that never become known to the police remain outside the statistics.

Cases must not be equated with victims or suspects. One case may include several victims or several suspects. The case-level knife-attack marker also does not reliably identify which individual suspect carried, threatened with or used the knife. In group offences, unarmed participants may be recorded alongside the person who handled the weapon. Treating every suspect in a marked case as a knife user would therefore be misleading.

A new catalogue called “type of weapon use” was intended to address this problem from 2025. It was designed to distinguish, at case and individual-suspect level, whether a knife had merely been carried, used to threaten or actually applied. This could have produced more reliable information about the people who handled knives and the manner in which they were used.

The extension failed in its first reporting year. During annual processing, one federal state reported quality-assurance problems. When producing the national tables, the BKA then identified inconsistencies across states and data combinations regarded as implausible or impossible. Once annual processing had been completed, the 2025 values could no longer be corrected retrospectively. The federal government therefore expects valid results no earlier than the 2026 reporting year.

This does not make the 2025 headline figures automatically wrong. It is nevertheless a significant data-quality failure. Public debate has long demanded exactly the detail that the new system was meant to provide: who actually handled the knife, whether the conduct involved carrying, threatening or applying it, and how those suspects were distributed across relevant characteristics. These questions still cannot be answered validly at national level for 2025.

It is particularly concerning that the inconsistencies became visible only while the national tables were being prepared. Uniform recording had been agreed in 2018, introduced in 2020 and harmonised over several years. Yet the system apparently lacked validation rules capable of reliably blocking impossible combinations during the reporting year. This is not evidence that the published core total was falsified. It does show that technical quality assurance did not match the political importance attached to the data.

There is also no national official dataset available for an independent cross-check, and no separate BKA situation report on knife crime is planned. Germany therefore lacks both a second national series for plausibility testing and a recurring in-depth analysis combining locations, injury outcomes, victim–offender relationships, situational context and differences in state recording practices.

How political decisions shape the statistics

Political influence on crime statistics does not automatically mean manipulation. It usually begins earlier: political institutions decide which phenomena should be measured, which criminal offences exist, which fields are mandatory and which figures receive prominence in public communication.

The knife-attack category illustrates this clearly. It was created following a decision by the Conference of Interior Ministers. Selected figures were published during the early harmonisation phase because the issue attracted particular political interest. Meeting that demand for information was legitimate, but it also produced a public sequence of numbers whose scope changed from year to year. Those different figures were later often treated as though they formed one continuous trend.

The chosen definition also shapes political perception. One overall total combines immediate threats with attacks that were carried out. This is permissible as a broad overview, but headlines can easily turn 29,243 “knife attacks” into the image of 29,243 physical assaults. Publishing threats, application without injury, injury and fatal outcome as equally prominent categories would present the same phenomenon more precisely.

Following the Solingen attack, Germany adopted a security package in October 2024. It expanded statutory restrictions on carrying knives at public events and on long-distance public transport and enabled additional weapons and knife-ban zones. The measures therefore applied during a substantial part of the 2025 reporting year.

The 2025 Police Crime Statistics cannot establish either the success or failure of these rules. A meaningful evaluation would require a credible before-and-after design, comparison areas, information about differences in implementation across federal states and a breakdown by locations actually covered by the new restrictions. There is also a conceptual distinction: merely carrying a knife is not recorded as a knife attack. More intensive controls may reveal more carrying violations without directly increasing the knife-attack total. A preventive effect on later offences is possible, but it cannot be identified from a single national annual change.

Changes to criminal law also influence measurement. The 2021 expansion of the threat offence widened the conduct subject to prosecution and created a break with earlier years. Such reforms may be justified, but they alter the statistical base. A rise following a legal expansion does not necessarily mean that the underlying behaviour increased to the same extent.

There is no substantiated evidence that the published 2025 core figures were politically manipulated. The documented problem concerns the technical recording of the new weapon-use catalogue, and the federal government states that no instructions were issued to publish or withhold particular data. The stronger criticism lies elsewhere: in the slow creation of a comparable series, the combination of very different forms of conduct, the continued absence of valid individual-suspect attribution and a public presentation that gives more weight to the total than to the shift within it.

The defensible reading of the 2025 figures is therefore narrower than many political interpretations. The overall number was almost stable, rising by 0.8 per cent. Knife attacks within violent crime fell by 6.1 per cent, while threats rose by 9.5 per cent. Only two fully comparable years exist for the overall total. The published national data cannot determine the causes of these movements, the effect of the latest knife restrictions or the characteristics of the individuals who actually used a knife. The statistics describe a substantial police-recorded field, but they still lack the depth required to support far-reaching political claims of cause and effect.