How to counter extremism: what works, what fails and what is unproven

The short answer. How to counter extremism is a question on which governments report plenty of activity, but independent evidence of what works is thin. Reviews in Britain, the Netherlands and France found little or mixed evidence of effectiveness. This guide sets out what each programme is, what the reviews found, and where the evidence is weakest.

Key facts #

• Britain's CONTEST strategy says agencies disrupted 39 late-stage terrorist plots between March 2017 and July 2023, and promises an evaluation programme to build "an evidence base on what works in counter-terrorism" (July 2023).¹
• William Shawcross's review found that of 30 evaluations of Prevent-funded projects only a third showed an overall positive impact, and 11 were ineffective or poor (8 February 2023).³
• A Dutch government review of counter-terrorism policy for 2018 to 2024 found little insight into its effectiveness and none into its cost-efficiency (19 November 2025).⁶
• France's Senate judged in March 2024 that the results of the 2021 "separatism" law were "peu probants" (not very convincing), yet called the temporary closure of places of worship a necessary measure (6 March 2024).⁷
• The first terrorism trial to come out of Austria's Operation Luxor raids of 9 November 2020 ended in an imam's final acquittal on 17 June 2026.⁸

How to counter extremism: what does the evidence show? #

Governments mostly report outputs, not outcomes. The UK's 2023 strategy promises a "systematic evaluation programme" to help build an evidence base on what works in counter-terrorism.¹ Australia's 2022 strategy lists "strengthening the evaluation of national counter-terrorism programs" as an action.⁹

Dutch reviewers add a caution: effect evaluation of counter-terrorism is internationally recognised as hard, because the intentions of would-be terrorists cannot be observed.⁶ A lack of proof is therefore not proof of failure.

What works to counter violent extremism? #

Three measures have some independent support. Each is narrower than the headlines suggest.

Disrupting plots. Governments count foiled plots: 39 in the UK to July 2023 and 31 major plots in Australia since 2014.¹² The tallies show operational output. They do not show what moved the threat up or down.

Fair policing. A study of Muslim Americans in New York in 2009 by Tom Tyler, Stephen Schulhofer and Aziz Huq found that the procedural justice of police activity was the primary factor shaping views of police legitimacy and willingness to cooperate.¹⁰ The EU Agency for Fundamental Rights found in 2024 that Muslims who saw their latest police stop as discriminatory profiling trusted the police at 4.2 on a 0 to 10 scale, against 6.8 for those not stopped.¹¹ Both studies measure trust, not terrorist tips received.

Supervised reintegration. The Dutch review found that the TER team, which manages the return to society of imprisoned extremists, was assessed as effective, with low reoffending and good cooperation between partners.⁶

Reoffending is low with or without such programmes, which makes success hard to read. Andrew Silke estimated in 2011 that "on average probably less than five percent" of terrorist prisoners are re-convicted for terrorist-related activity.¹²

Does the UK's Prevent strategy work? #

Prevent is the strand of CONTEST that tries to stop people becoming terrorists, and public bodies have a statutory duty to take part.¹ In the year to September 2025 it received 10,293 referrals, and 18 per cent were adopted as Channel cases. Where a concern was recorded, 56 per cent had no identified ideology, 20 per cent were extreme right-wing and 8 per cent were Islamist.⁵

Shawcross, the independent reviewer, praised Prevent's "sophisticated and impressive" architecture and rejected the "caricature" of it as "an authoritarian and thinly veiled means of persecuting British Muslims". He also found "a double standard": an expansive approach to the extreme right and a narrow one to Islamism. He noted that 80 per cent of counter-terrorism police investigations were Islamist, against only 22 per cent of Prevent referrals in 2020-21.³

On results, he examined 30 individual evaluations. Only a third showed an overall positive impact, and 11 were ineffective or poor. Five projects reduced participants' support for free speech that offends others.³ The government accepted his recommendation that Prevent's first objective should be to "tackle the ideological causes of terrorism".⁴ He also wrote that "large numbers of people" were referred "who do not need to be in the scheme", yet judged that the Channel process "is of great value".³ For the signs of radicalisation, see the radicalisation guide.

What is the Aarhus model, and did it work? #

The Aarhus model is the Danish city's way of handling young people drifting towards violent extremism and returning foreign fighters, through mentoring and exit programmes run by the municipality and East Jutland Police. The EU's Radicalisation Awareness Network describes it but gives no outcome figures.¹³

The reported results are encouraging. NPR counted 34 people who left Aarhus for Syria from 2012: six killed, ten still there and 18 back home. Only one person left in 2015, "even when traffic from the rest of Europe was spiking".¹⁴ A French Senate report gave "some 36" departures up to 2015.¹⁵

The limits are just as clear. The same report found no statistics on results, and critics had called the model naive and said it risked "collusion" with radicalised people.¹⁵ Preben Bertelsen of Aarhus University notes that Denmark's strategy also includes prosecution, passport confiscation and security-service action.¹⁶ The model supplements policing. It does not replace it.

Do deradicalisation programmes work? #

Deradicalisation (US spelling "deradicalization") means changing beliefs. Disengagement means ending violence. Silke found no thorough evaluation of jihadist programmes in 2011, and argued that they generally "work" only with "soft core" militants.¹²

France's residential centre at Pontourny is the clearest failure. Opened in September 2016 as the first of a planned 13, it had 25 places and never held more than nine people at once. No resident completed the programme. A Senate report of July 2017 cited its "absence totale de résultat", put its running cost at about 2.5 million euros a year while it held nobody, and proposed ending it. The report preferred support in the person's own community, as at Aarhus.¹⁵

Release is the other weak point. Usman Khan, convicted of terrorism offences in 2012, was released in December 2018 under licence with a GPS tag, multi-agency management and Home Office mentors, despite being assessed as high risk. On 29 November 2019 he killed two people at Fishmongers' Hall.¹⁷ The coroner nonetheless said rehabilitation programmes such as Learning Together can "help to move people away from offending behaviour", if run "in a safe and intelligent manner".¹⁷

What did France's 2021 separatism law achieve? #

Law no. 2021-1109 of 24 August 2021 answers what President Macron called "le séparatisme islamiste", a "conscious, theorised, politico-religious project", in his Les Mureaux speech of 2 October 2020. He also warned against stigmatising all Muslims.¹⁸ The Constitutional Council struck down three provisions on 13 August 2021 and upheld others.¹⁹

The law lets prefects close places of worship temporarily where statements or activities incite hatred or violence, tightens the dissolution of associations, and makes public subsidies depend on signing a "republican commitment contract".⁷ Nearly three years later the Senate said the balance sheet was "far from conclusive". Its main negative finding concerned the commitment contract, which it called "more an administrative formality than a real commitment". Senators found no subsidy refused under it and counted four withdrawals in two and a half years, most from militant rather than separatist bodies. A truly separatist association can avoid subsidies altogether and pass "under the radar".⁷

Other findings were positive. The Senate called temporary closure of places of worship "necessary", and the Interior Minister reported 18 closures from the end of the Covid emergency to February 2024. It called the balance of the dissolution reforms "rather positive", though the list included ultra-right and ultra-left groupuscules and an annulled dissolution of Les Soulèvements de la Terre. The foreign-funding declaration rules "seem to have produced their effects": reports of suspicious transactions about worship fell by 30 per cent.⁷ Neither the Senate nor the sources here show a fall in attacks. The French case and the Brotherhood networks it targets are covered in the Europe page and the Muslim Brotherhood guide.

What happened after Austria's Operation Luxor? #

Austria went after what its intelligence service calls "legalistic Islamism": Islamists who pursue their aims without violence inside the democratic legal order. On 9 November 2020, a week after the Vienna attack, police searched homes and premises of people suspected of being active in that field and close to the Muslim Brotherhood.²⁰ Authorities spoke of 70 suspects, on suspicion of terrorism financing and forming a terrorist organisation.⁸ The courts then narrowed the case:

• August 2021: the Higher Regional Court in Graz upheld complaints from people whose homes were searched. Its spokesperson said the raids "were not lawful" and that the Muslim Brotherhood is not considered a terrorist organisation in Austria (as reported by Anadolu Agency).²¹
• January 2023: the same court ended the investigation of the political scientist Farid Hafez, finding "no substance" for the suspicion. Proceedings against 31 suspects had by then been discontinued.²²
• 17 June 2026: a Vienna court acquitted an imam of terrorist-organisation membership and incitement, finding that a video and three sermons were covered by freedom of opinion and religion. The verdict is final.⁸

Proceedings against 13 suspects were still open in June 2026, and the Interior Minister said the investigations were not over.⁸ An acquittal does not settle the ideological question: Austria's 2020 intelligence report states that the Brotherhood's goal of an Islamist state is "not compatible" with democratic principles.²⁰

What works against Islamism, as opposed to violence? #

The hard question is the non-violent end of the spectrum. Shawcross argued that Prevent must cover "non-violent Islamist narratives and networks" that help create an environment conducive to terrorism.³ The Austrian and French cases show the difficulty, and the sources here show no fall in Islamist violence as a result in either country.

The site's own recommendations are in the Islamism guide on stopping Islamist terrorism and on how free societies stop Islamism: prosecuting incitement, one law for everyone, defending free speech and apostates, and backing Muslims who reject Islamism. The ideas to be countered are in the guide to the Islamist ideologues.

What do Australia's programmes show? #

Australia's intervention programmes are voluntary. The Commonwealth and the states fund the Living Safe Together Intervention Program (A$19.6 million, October 2022 to June 2025), succeeded by the National Support and Intervention Program (September 2025 to June 2028).²³ The 2022 strategy says such programmes "reduce the risk of violent extremist incidents".⁹ That is a statement of purpose, not a result.

The independent evaluation in the sources is state-level. ACIL Allen's October 2019 evaluation of New South Wales's A$47 million countering violent extremism program found that it reached nearly 1.5 million people, but that diversion and disengagement were hard to measure. It warned that "connections with CT securitises CVE programs which risks community engagement and confidence", yet found the program "relatively well received by community stakeholders when compared to other CVE initiatives".²⁴ See the guide to Islamism in Australia for the threat these programmes address.

What cuts the other way? #

• Muslims are partners, not only suspects. ACIL Allen found a service built on "support, not report" to be "valuable and well received by communities".²⁴
• Some hard tools passed legal tests. The Council of State held in June 2023 that France's commitment contract did not violate freedom of association or expression.⁷
• These programmes are not only about Islamism. In the Prevent figures above, extreme right-wing referrals outnumber Islamist ones.⁵

Questions people ask #

How can extremism be stopped? #

No single policy has been proven to stop it. Governments combine intelligence-led disruption of plots, prosecution of incitement and violence, support schemes for people drifting towards extremism, supervised release of offenders and policing that Muslims see as fair. Evaluations are thin, so each measure should be tested and published rather than assumed to work.

What is the Prevent strategy? #

Prevent is the part of Britain's CONTEST counter-terrorism strategy that aims to stop people becoming terrorists. Public bodies have a statutory duty to take part, and a panel decides whether referred people receive Channel support. The 2023 Shawcross review praised its structure but criticised an uneven focus on Islamism and weak evidence for many funded projects.

Do deradicalisation programmes work? #

The evidence is not strong enough to say. Terrorist prisoners reoffend at low rates with or without programmes, and no thorough evaluation of jihadist programmes was available in 2011. France's residential centre at Pontourny, opened in September 2016, was empty by February 2017. Support in the person's own community, as in Aarhus, looks more promising.

What is the Aarhus model? #

The Aarhus model is a Danish approach in which the police and city authorities offer mentoring and support to young people at risk of extremism and to returning foreign fighters. NPR counted 34 departures from the city to Syria since 2012 but only one in 2015. Senators who visited found no statistics on results, so its effect is unproven.

What works against Islamism? #

Tools aimed at violence, such as surveillance, prosecution and supervised release, are well established legally. Tools aimed at non-violent Islamism have a weaker record. Austria's raids on alleged Muslim Brotherhood supporters were ruled unlawful on complaint and the first terrorism trial ended in acquittal, while France's 2021 law drew a mixed verdict from the Senate.

How do you counter violent extremism without alienating Muslims? #

Evidence from New South Wales and the EU points to keeping community programmes away from security branding, treating people fairly when police stop them, and offering support rather than only reporting. Muslims in the EU trusted police less when they thought a stop involved profiling.

Related guides #

• Islamist radicalisation: how people are drawn in, the signs, and where to report.
• Islamism explained: what an Islamist is, and how free societies stop Islamism.
• Islamist terrorism in the UK: attacks, Prevent and the British response.
• Islamism in Australia: the threat, ASIO and the Bondi attack.
• The Muslim Brotherhood: goals, entryism and bans.
• Islamist terrorist attacks in Europe: the attacks, year by year.

Cite as: The Analyst, "How to counter extremism: what works and what fails", Fighting Islamists, updated 13 October 2026. https://fightingislamists.com/counter-extremism.html

Next: France 2027 election →