The term “knife attack” is frequently used in news reports, political debate and public statistics as though it referred to a clearly defined criminal offence. In fact, the meaning is more differentiated. There is no separate criminal offence called “knife attack” under German criminal law. In the German Police Crime Statistics (Polizeiliche Kriminalstatistik, PKS), it is instead an additional characteristic assigned to an already recorded criminal case.
This distinction is essential when interpreting the figures. A case may, for example, be recorded as robbery, dangerous bodily harm, threat or homicide and additionally be marked with the phenomenon “knife attack”. The statistical category does not require an actual stab or cut injury. An immediate threat against a person involving a knife may also meet the criteria.
Case numbers, victim numbers and suspect numbers must likewise be distinguished from one another. A single case may involve several victims or several suspects. In offences committed jointly by several people, not every suspect necessarily used a knife. A proper interpretation of knife-attack statistics therefore requires an understanding of what the PKS actually counts and what information the characteristic “knife attack” contains.
One term, several different levels
Criminal law and crime statistics describe the same real-world incident from different perspectives. The actual event, its criminal-law classification and its statistical recording must therefore not be treated as identical.
“Knife attack” is not a separate criminal offence
German criminal law contains no separate offence called “knife attack”. Instead, the question under criminal law is what specific conduct occurred and which criminal offence that conduct may constitute.
An attack involving a knife may, for example, occur in connection with dangerous bodily harm, robbery, extortion by force or threat, or an attempted or completed homicide offence. Threat offences and other criminal offences may also be relevant depending on the circumstances. Important factors include the precise sequence of events, the intention of the person involved, the way in which the object was used and any resulting injuries.
The knife itself may have particular legal significance within these offences. Depending on its characteristics and use, it may, for example, be classified as a weapon or dangerous instrument, or it may serve as the means by which force is used or threatened. This can affect the criminal-law assessment.
This legal level must, however, be distinguished from the statistical term “knife attack”. The incident is investigated in relation to one or more specific criminal offences and may additionally receive the corresponding phenomenon marker in the PKS.
Key distinction: The offence code describes which criminal offence was recorded by the police. The phenomenon “knife attack” additionally describes a particular characteristic of the incident.
What the Police Crime Statistics record
The PKS itself is not a statistic of final criminal convictions. It is a police output statistic. As a general rule, the information available once police processing of the case has been completed is recorded. Any subsequent assessment by the public prosecutor’s office or a court must be distinguished from this.
A person listed as a suspect in the PKS is therefore not automatically a person who has been finally convicted. Proceedings may later be discontinued, an allegation may be legally reclassified or an accused person may be acquitted. Police crime statistics and judicial prosecution statistics are based on different recording dates and counting methods.
At the same time, not every unconfirmed allegation becomes a PKS case. Statistical recording requires sufficiently specific and verified indications concerning the incident. These include, in particular, information about the suspected offence, the place of the act and the time of the act or at least the relevant time period.
This system is particularly important in cases involving knives. An initial police statement may, for example, refer only to a “pointed object”, while the precise instrument is established later during the investigation. A press release, the actual state of the investigation and the subsequent PKS recording therefore need not reflect the same level of information.
How the nationwide recording of knife attacks developed
An offence code alone does not show whether a knife was used during a crime. Dangerous bodily harm may, for example, be committed with a knife, by kicking or punching, or with another object. Likewise, the statistical recording of a robbery alone does not identify the instrument used.
Against this background, the Conference of Interior Ministers agreed in June 2018 that knife attacks should in future be recorded according to a uniform nationwide statistical system. Since 1 January 2020, the phenomenon “knife attack” has been recorded throughout Germany as additional information relating to a case in the PKS.
The introduction of the data field did not, however, mean that unrestricted and directly comparable national data were immediately available. Newly introduced PKS catalogues initially pass through a phase in which, among other things, recording practices and data quality are assessed. According to the German Federal Government, valid nationwide data for the catalogues introduced at that time, including the phenomenon “knife attack”, have been available from reporting year 2024 onwards.
This development is important for longer-term comparisons. The fact that a characteristic has technically been collected since 2020 does not automatically mean that every value from the first year of recording can be compared without methodological restrictions with later national figures.
What the PKS defines as a knife attack
According to the guidelines governing the German Police Crime Statistics, knife attacks are acts in which an attack with a knife is immediately threatened against or carried out against a person. Merely carrying a knife is expressly insufficient.
Several distinctions follow from this definition. First, the conduct must be directed against a person. Finding a knife, possessing one or having one seized during a police check does not by itself constitute a knife attack. A violation of rules governing the carrying of knives also does not automatically fulfil the criterion.
Physical injury, on the other hand, is not required. An immediate threat of attack against a person involving a knife may already satisfy the phenomenon marker. The statistical category is therefore broader than the common assumption that a knife attack necessarily requires a stabbing or cutting injury.
The total figure consequently combines different courses of events and different offences. A threat during a robbery, a stabbing movement that is successfully avoided, dangerous bodily harm resulting in an actual cut, and attempted homicide may all carry the same additional phenomenon marker.
What the PKS understands by a knife
The instrument used is also defined more precisely for statistical purposes. The PKS guidelines include various cutting instruments consisting of a blade and a handle. These include pocket and folding knives, kitchen and household knives, utility knives, hunting knives, scalpels, daggers, switchblades, stilettos, straight razors and machetes.
The statistical definition is therefore not limited to objects specifically manufactured as weapons. An ordinary kitchen knife may qualify as the instrument used in a knife attack if it is used or immediately threatened against another person in the manner described by the PKS.
Conversely, not every sharp, pointed or bladed object is regarded as a knife within the PKS. The guidelines exclude, among other things, axes, sabres, swords, épées, foils, sickles and scythes from the knife definition.
Statistical distinction: A machete is included in the knife definition used by the PKS. A sword, by contrast, is expressly excluded. An attack involving a blade is therefore not automatically a “knife attack” under the PKS definition.
Threat, attack, attempt and injury are different characteristics
One of the most common sources of misinterpretation is the mixing of several different pieces of information. The knife-attack definition initially distinguishes between an immediately threatened attack and an attack that is carried out. Whether the underlying criminal offence was completed or merely attempted is a separate issue. Any actual injury must again be considered independently.
If, during a robbery, a victim is immediately threatened with a knife, the case may receive the phenomenon marker “knife attack” even though no stabbing, cutting or physical contact occurs. “Knife attack” and “knife injury” are therefore not synonymous.
Likewise, an attack that is carried out does not necessarily result in injury. A person may attempt to stab another person while the victim avoids or blocks the attack. The attack itself and its physical consequences are separate pieces of information.
The term “attempt”, by contrast, refers to the underlying criminal offence. The PKS also records punishable attempts where the attempted form of the offence is criminalised. The phenomenon “knife attack” remains an additional characteristic of that case.
An attempted homicide can therefore simultaneously involve a knife attack that was actually carried out. If a person attempts to stab another with intent to kill but misses, the homicide offence may remain at the attempt stage even though the knife attack itself was physically carried out.
The 2025 PKS illustrates this distinction. Within violent crime, 14,787 cases were recorded with the phenomenon “knife attack”. Of these, the underlying offences were recorded as completed in 10,819 cases and attempted in 3,968 cases. The proportion of knife attacks among attempted violent offences was 11.5 percent, higher than the 6.1 percent recorded among completed violent offences.
These figures expressly do not mean that 10,819 people were successfully injured with a knife and that 3,968 knife attacks failed. “Completed” and “attempted” refer to the underlying criminal offence.
What the 2025 PKS actually reports
For reporting year 2025, the nationwide Police Crime Statistics record a total of 29,243 criminal offences with the phenomenon “knife attack”. In reporting year 2024, the corresponding figure was 29,014. The total therefore increased by 229 cases, or 0.8 percent.
Behind this almost stable overall figure, however, are different developments. In 2025, 50.6 percent of knife attacks were assigned to the statistical category of violent crime, 47.0 percent to threat offences and 2.4 percent to other offences. In 2024, the corresponding distribution had been 54.3 percent violent crime, 43.3 percent threat offences and 2.4 percent other offences.
Violent crime as a whole also declined in 2025. The number fell from 217,277 to 212,335 cases, a reduction of 2.3 percent. Knife attacks within this category decreased more sharply, by 6.1 percent. Their share of all violent-crime cases was 7.0 percent in 2025 compared with 7.2 percent in the previous year.
The term “violent crime” is itself a precisely defined statistical category. The PKS aggregate includes, among other offences, murder and manslaughter, robbery, dangerous and serious bodily harm and other specified offences. It is not identical to the everyday understanding of everything that may be considered “violent”.
It would therefore also be incorrect to conclude that knife attacks outside this aggregate were “non-violent” incidents. A substantial proportion consists of threat offences in which a knife may have been immediately threatened against another person.
Significant differences between offence groups
Knife attacks within violent crime do not form a uniform category either. Absolute case numbers and the respective proportion of knife attacks among all cases within an offence group can differ considerably.
For murder, manslaughter and killing at the victim’s request, 994 cases carrying the phenomenon knife attack were recorded in 2025. A knife therefore played a corresponding role in 40.5 percent of the cases recorded within this statistical category. Compared with 922 cases in the previous year, this represents an increase of 7.8 percent.
For robbery offences, 4,441 cases involving knife attacks were recorded. They accounted for 11.1 percent of all robbery cases. Compared with 4,768 cases in 2024, the figure declined by 6.9 percent.
The highest absolute figure among the listed categories occurred in dangerous and serious bodily harm: 9,227 cases were recorded with the phenomenon knife attack. This corresponded to 6.0 percent of all cases in this offence group. In 2024 there had been 9,917 such cases, resulting in a decline of 7.0 percent.
These differences show why both absolute numbers and relative proportions are required. Homicide offences involve far fewer knife-attack cases in absolute terms than bodily-harm offences, while the proportion of knife attacks within homicide offences is considerably higher.
A threat offence and threatening someone with a knife are not the same thing
An especially important statistical distinction concerns the underlying criminal offence and the specific way in which the knife was used. The fact that 47.0 percent of knife attacks in 2025 were assigned to the offence category of threats does not mean that knives were used as a threat only in these cases.
One example is a robbery committed while holding a knife on the victim. The knife may be used exclusively as a means of threat throughout the incident. The underlying criminal offence nevertheless remains robbery and is therefore statistically part of violent crime.
Conversely, the category “threat” describes cases in which the underlying offence was classified accordingly. The offence category therefore answers a different question from the specific manner in which the knife was used.
The proportion of threat offences must therefore not be described as the proportion of all knife attacks in which the offender “only threatened” with the knife. A separate PKS catalogue was introduced to record more precisely how often knives were carried, used to threaten or actually applied.
Case, victim and suspect are different counting units
Another major source of error is the equation of statistical cases with people. A PKS case is not a person. A victim is not a case. A suspect is not necessarily the person who used the knife.
For victim recording, the PKS counts natural persons against whom a punishable act is directly directed, provided that the relevant offence code provides for victim recording. A case can therefore involve several victims. The same person may also be recorded as a victim in different cases.
The 29,243 criminal offences carrying the phenomenon knife attack therefore do not equate to either 29,243 victims or 29,243 injured persons. A victim may only have been threatened, may have avoided an attack or may actually have suffered an injury.
Why suspects are not automatically knife users
This distinction is particularly relevant for offences committed jointly by several people. According to the PKS guidelines, where a jointly committed offence involves different instruments, it is sufficient for at least one suspect to have committed a knife attack for the corresponding case marker to be recorded.
If, for example, three people jointly commit a robbery and only one of them threatens the victim with a knife, the case may carry the phenomenon “knife attack”. The other two suspects nevertheless belong to the same case even though no personal knife use need be attributed to them.
The characteristics of all suspects in such cases therefore cannot automatically be presented as characteristics of the actual knife users. This is particularly relevant for analyses relating to age, sex or nationality.
In addition, suspect statistics follow their own counting logic. The PKS uses what is known as a real-suspect counting method, intended to prevent the same person from being counted repeatedly as though they were different individuals merely because they were involved in several similar offences within the relevant statistical category.
Several cases may therefore involve the same suspect. Conversely, a single case may involve several suspects. Case numbers and suspect numbers must consequently not be equated directly.
Methodologically invalid: Treating all suspects in a case carrying the phenomenon “knife attack” as “knife users”, “knife offenders” or “stabbers” when the individual use of the knife has not been established.
The new catalogue “type of weapon use”
To address this information gap, an additional catalogue on the “type of weapon use” was introduced for reporting year 2025. It is intended to record whether a knife was carried, used to threaten or actually applied.
The new catalogue is designed both for case-based recording and, in solved cases, for individual suspects. This is intended to make it possible in future to determine more precisely which person actually carried a knife, threatened someone with it or used it.
This additional information answers a different question from the phenomenon marker “knife attack”. The phenomenon identifies the case. The type of weapon use describes more precisely how the knife was used and to which suspect that use can be attributed.
For reporting year 2025, however, these new data were not yet valid nationwide. According to the Federal Government, the annual processing revealed inconsistencies between the federal states as well as partly implausible or impossible combinations of data. Once the annual PKS processing had been completed, the differing recording practices could no longer be corrected retrospectively.
Important: This does not mean that the 2025 PKS knife-attack figures as a whole are invalid. The Federal Government expressly states that the lack of validity applies only to the new values in the “type of weapon use” catalogue. The published data relating to the phenomenon “knife attack” itself are regarded as valid.
For reporting year 2026, the Federal Government considers it realistic that valid data from this new person-specific recording system will become available for the first time. This should make certain statements about actual knife users at national level considerably more precise.
A special feature of “threatened with a knife”
The PKS guidelines contain a particular rule for the new recording of weapon use. “Threatened with a knife” is recorded where at least one victim subjectively feels threatened with a knife. This can also apply where the victim merely assumes that the object involved is a knife.
This rule belongs to the new “type of weapon use” catalogue. It must not automatically be equated with the general definition of the phenomenon knife attack or with a later objective determination of the actual instrument used.
Professional analysis must therefore make clear which information is actually available: the victim’s perception, a police determination regarding the object, the knife-attack case marker or the specific type of weapon use.
Merely carrying a knife is not a knife attack
The official definition contains a clear negative distinction: merely carrying a knife is insufficient for recording a knife attack.
Different areas that are often discussed collectively under the heading of “knife crime” must therefore be separated. A violation of legal restrictions on carrying knives, a knife discovered during a police check and an actual knife attack directed against another person are statistically different circumstances.
A person may be carrying a prohibited knife during a police check without a knife attack having taken place. Likewise, a knife may have been carried during another offence without being immediately threatened or used against another person.
The distinction in the new catalogue between “carried”, “used to threaten” and “applied” makes particularly clear that the mere presence of a knife is not equivalent to a knife attack.
The PKS reporting year is not necessarily the year of the offence
The timing of cases is also important when interpreting annual figures. The PKS is an output statistic. A case is generally included statistically in the period in which police processing has been completed and the case has been submitted accordingly.
An offence may therefore occur at the end of one year and only be included in the following PKS reporting year. Conversely, a reporting year may contain cases whose actual offence date was in the previous year.
The shortened statement “29,243 knife attacks in 2025” can therefore be misleading. The statistically more precise wording is: In reporting year 2025, 29,243 criminal offences carrying the phenomenon “knife attack” were recorded in the PKS.
This distinction becomes especially important when PKS figures are compared with databases that assign incidents according to the actual date of the event.
The PKS describes recorded crime
The Police Crime Statistics do not represent every crime that actually occurred. A prerequisite for statistical recording is first of all that the incident became known to the police.
Offences that are not reported and do not otherwise come to police attention remain outside the PKS. This unrecorded area is commonly referred to as the dark field of crime.
Changes in recorded case numbers therefore cannot be interpreted exclusively as changes in the actual level of crime. Reporting behaviour, police control and investigative activity, legislative changes, changes in statistical definitions and the quality of newly introduced data fields may also influence the recorded figures.
The PKS nevertheless remains the central nationwide source of data on crime recorded by the police. Its figures must be interpreted within the methodological limits described by the Federal Criminal Police Office itself.
Common misinterpretations of knife-attack statistics
The different statistical levels give rise to several statements that may initially appear plausible but are not supported by the available data.
“29,243 knife attacks mean 29,243 injured people.”
A case count is not a victim count. In addition, an immediate threat alone may satisfy the phenomenon marker.
“Every knife attack is a stabbing.”
The official definition also includes immediate threats. An actual stabbing movement is not required.
“All suspects in a knife-attack case used a knife.”
In jointly committed offences, suspects may belong to the same case even though only one participant used the knife.
“47 percent threat offences means that knives were used to threaten in 47 percent of cases.”
The offence category describes the underlying criminal offence. A knife may also have been used solely as a means of threat during a robbery.
“The 2025 knife-attack data are invalid.”
The newly introduced detailed values in the “type of weapon use” catalogue were not valid. The published data relating to the phenomenon knife attack itself are regarded as valid by the authorities.
“All knife attacks in reporting year 2025 were committed in 2025.”
Because the PKS is an output statistic, the date of the offence and the reporting year may differ.
Distinction from MesserMonitor’s incident documentation
MesserMonitor follows a different methodological approach from the German Police Crime Statistics and does not attempt to reproduce their case-counting system. While the PKS records police-processed criminal cases according to uniform statistical rules, MesserMonitor documents publicly substantiated incidents on the basis of publicly accessible and verifiable sources.
Several public reports may be assigned to the same actual incident. Later follow-up reports can supplement or correct existing information. Wherever possible, MesserMonitor assigns incidents according to the actual date of the event rather than the date on which a police case was statistically completed.
The substantive scope is also not completely identical. Under its published inclusion criteria, MesserMonitor covers not only knives but also other substantiated stabbing and bladed weapons. An incident involving a sword may therefore fall within MesserMonitor’s scope even though a sword is not classified as a knife under the PKS definition.
Sources may include publicly accessible statements from police and investigative authorities, other official publications, reliable regional and national media reports and publicly verifiable follow-up reports. Several sources may be used to supplement and plausibility-check an incident. A search result or an unverified republication is not, by itself, sufficient evidence.
Because of these different recording principles, PKS case numbers and MesserMonitor incident numbers cannot be equated directly. Different counting units, time references, definitions of relevant instruments, source availability and inclusion criteria can systematically lead to different results.
The German Police Crime Statistics remain the authoritative official statistics for crime recorded by the police. MesserMonitor complements this perspective with event-based documentation of publicly accessible information according to its own published methodology.
What can be derived from the data
The available PKS data permit reliable statements about how many criminal offences within a reporting year were marked with the phenomenon knife attack, to which published offence categories these cases were assigned and how the figures developed compared with comparable data from the previous year.
For 2025, for example, it can be established that the overall number increased by 0.8 percent compared with 2024 to 29,243 cases, while the number of knife attacks within violent crime fell by 6.1 percent to 14,787 cases. At the same time, the proportion assigned to the offence category of threats increased. The almost unchanged overall total therefore conceals different developments within the offence groups it contains.
The overall figure alone does not establish how many people were actually injured by knives, how many individual stabbing or cutting injuries occurred or which specific suspect used the knife in every case. Statements about personal characteristics of actual knife users likewise require data in which the use of the knife can be reliably attributed to a particular person.
When interpreting knife-attack statistics, it must therefore always be clear which statistical unit, definition and data basis underlie a statement. The case marker, underlying criminal offence, type of use, number of victims and number of suspects describe different aspects of the same field and must not be treated as interchangeable.
Sources
- Federal Criminal Police Office (BKA): Guidelines for the German Police Crime Statistics 2025
- Federal Criminal Police Office / Conference of Interior Ministers: Police Crime Statistics 2025 – Selected Figures at a Glance
- Federal Criminal Police Office: Significance, Content and Informative Value of the Police Crime Statistics
- German Bundestag, Printed Paper 21/5507: Status of the Recording of Knife Crime at Federal Level
