Skip to content

Produced by the Sentencing Academy

Restraining Orders

Peer-reviewed research on the effectiveness and impact of sentences, orders, and requirements.

Melissa Hamilton, University of Surrey

Published: 17 Sep 2025
Last updated: Sep 2025

Summary

This Bulletin examines restraining orders that courts in England and Wales may impose at the point of sentencing—formerly under the Protection from Harassment Act 1997 and now under the Sentencing Act 2020. These orders are one of several forms of protective order, but are distinctive in being issued alongside a criminal sentence, with conditions tailored to prevent further harassment or threats from the convicted person.

There is currently no national reporting on how many such orders are issued each year, nor any restraining order-specific reoffending statistics. As a result, effectiveness cannot be assessed using the standard recidivism measures typically relied upon in effectiveness reviews. Instead, this Bulletin draws on proxy indicators—such as the scale of reported breaches, prosecution volumes, and sentencing outcomes. These suggest that breaches are relatively common, that enforcement is mixed (with some breaches leading to prosecution and others to no action), and that there is a trend toward greater use and length of custodial sentences for those convicted of breach.

Overall, restraining orders issued at sentencing remain a key tool for victim protection. Yet, without fuller and more consistent national data, it is not possible to determine whether they reliably deter reoffending or consistently deliver a sufficient level of safety for their intended beneficiaries.

Restraining orders are used as a tool for victim protection

Restraining orders issued at the point of sentencing are one of the tools courts use to protect individuals from harassment or threats. While these orders are preventative in nature, their effectiveness depends both on offender compliance and on consistent enforcement when breaches occur.

This Bulletin examines the use and effectiveness of restraining orders imposed on convicted defendants at sentencing. Drawing on national statistics, freedom of information disclosures, and research published in England and Wales from 2011 forward, it reviews how often these orders are used, the scale of reported breaches, and how courts respond when individuals are convicted. The analysis considers enforcement patterns, sentencing trends, and stakeholder confidence in the system, alongside the significant data gaps that hinder a full assessment of impact.

There are no publicly available reoffending statistics that isolate breaches of restraining orders as a distinct offence category. This absence means that effectiveness cannot be measured using the same core recidivism indicators applied in similar reviews. Instead, the Bulletin relies on proxy measures—such as prosecution and conviction rates for breaches of orders, practitioner perspectives, and targeted studies—to build the best available picture.

National statistics do not track the total number of restraining orders in England & Wales

In England and Wales, courts have statutory authority to impose restraining orders at the point of sentencing—whether or not a custodial sentence is imposed—to protect individuals from harassment or conduct that causes fear of violence. The primary legal basis for such orders was formerly found in Section 5 of the Protection from Harassment Act 1997 (PHA), now consolidated within Section 363 of the Sentencing Act 2020.1 These provisions empower criminal courts to issue restraining orders upon conviction for any offence, not solely harassment-related crimes, where the court considers it necessary to protect a person from harassment or threat by the convicted individual.

Restraining orders are a type of ancillary order. This means they are imposed in addition to, rather than in place of, a substantive sentence, such as imprisonment or a Community Order. Restraining orders are preventative rather than punitive in intent, functioning as civil-type measures within the criminal process to proactively reduce the risk of future harm.

The scope of a restraining order can be broad, such as including prohibitions on contact with named individuals, exclusion from specified areas, or bespoke restrictions addressing the risks posed by the convicted person. Orders may be time-limited or indefinite (remaining in effect until further court order). A breach of a restraining order without reasonable excuse is a criminal offence, punishable by up to five years’ imprisonment and/or a fine.

There are currently no publicly available national statistics on the total number of restraining orders issued at sentencing. Nonetheless, the overall numbers from an earlier period provide some insight. Bates and Hester (2020) collected data from several sources, reporting that about 18,800 restraining orders were issued per year, on average, across calendar years 2011-2017.2

The deterrent effect of restraining orders is unclear, as breaches of these orders are common.

A common starting point for evaluating sentencing effectiveness is to examine reoffending rates by offence and type of sentence given (Gormley, Hamilton, and Belton, 2022). However, no publicly available reoffending statistics in England and Wales isolate breaches of restraining orders as a distinct category. In the Ministry of Justice’s proven reoffending datasets, these breaches are subsumed within a broader grouping referred to as “Violence Against the Person” (Ministry of Justice, 2024). Further, the reoffending dataset does not separately track restraining orders as a type of sentence, likely because it is considered an ancillary order rather than a main form of punishment. In lieu of using reoffending as a major focus here, then, the evaluation of the effectiveness of restraining orders must rely on proxy measures—such as information on breaches of the orders, qualitative accounts from practitioners, and targeted studies.

Restraining orders are more likely to be effective when the prospects of enforcement and sanction for breach are consistent and credible (Bates and Hester, 2020). The Domestic Abuse Commissioner for England and Wales has warned that ‘without robust enforcement, protective measures [such as restraining orders] are rendered functionally futile. Perpetrators are emboldened to continue their abuse without consequence, while survivors are put at further risk by giving them a false sense of security’ (Jacobs, 2025: 46). Her 2025 report noted that failure to arrest for breaches is common, and that enforcement practices vary significantly between police forces for a variety of reasons. Oppenheim (2024) also relays anecdotal stories from victims and experts suggesting that breaches are often not reported at all, due, in part, to a lack of confidence that police action will follow.

Concerns from victim advocacy groups reinforce this point. A government report noted that Women’s Aid criticised protective orders generally (of which restraining orders are a type) for failing to provide adequate protection, citing poor monitoring and inconsistent enforcement that allowed perpetrators to “act with impunity” (Brader, 2024). The report also noted that the Centre for Women’s Justice similarly warned that without a radical improvement in implementation, such orders would have limited impact.

Because there are no national statistics on the overall frequency of restraining order breaches, available data instead provide partial snapshots at different stages of the enforcement process. These sources are not directly comparable but, taken together, they offer a sequential view—from police charging decisions to prosecutions and final court outcomes.

At the first stage, freedom of information disclosures offer some insight into police charging decisions in restraining order breach cases, though the ad hoc nature of such disclosures means the figures span different time periods and are not directly comparable. Drawing on data from 21 police forces, Oppenheim (2024) reported that just over half (52%) of reported breaches in the year 2023 resulted in criminal charges. Looking at individual forces highlights further variation. For example, one police force reported that between 2018 and 2023, 45% of breaches resulted in a charge, 29% were not pursued due to evidential difficulties, 18% were closed because the victim declined or withdrew support, 3% were discontinued on public interest grounds, and the remainder were closed for other reasons (Cleveland Police, 2024).3 In contrast, another police force charged 36% of restraining order breaches between 2020 and 2023, without listing the reasons why almost two-thirds were closed without charge (West Midlands Police, 2024).4 These figures illustrate variation across forces, suggesting issues with availability of evidence, victim engagement, and local practices in handling restraining order breaches. It is also important to note that the figures presented here relate only to police charging decisions and do not show how many charges were ultimately prosecuted or led to convictions.

At the next stage, Bates and Hester (2020), analysing data collected from several sources nationally, found that almost 40% of restraining orders issued during the period 2011-2017 led to prosecutions for breach—amounting to more than 57,000 prosecutions across those years. Consequently, a significant number of restraining orders are prosecuted for noncompliance.

At the court stage, the Ministry of Justice provides the most complete national statistics for the final stage of the process, reporting the number of defendants prosecuted, convicted, and sentenced for breaches of restraining orders. These data, covering 2017 to 2024, are presented in Table 1 and capture only cases that were formally prosecuted.

Table 1: Volume of prosecutions and custody rates for breach of restraining orders5

20172018201920202021202220232024
Prosecuted10,0879,7898,3877,1416,3285,9605,9736,294
Convicted9,3699,7017,8116,7155,8655,5435,4505,675
Sentenced9,3639,0637,8386,7995,8235,6155,3185,680
Custody Rate39%39%42%47%46%51%51%52%

Table 1 shows a steady decline in the numbers of prosecutions for breaches of restraining orders, falling from 10,087 (year 2017) to 6,294 (2024)—an overall reduction of nearly 40%. Some commentators interpret this trend as evidence of a fading interest in enforcing these orders (Mistry, 2022). However, without accurate data during those years on how many restraining orders were issued or how many breaches actually occurred, it is not possible to say whether prosecutions have in fact become less likely. If the number of restraining orders issued, or the volume of actual breaches, also declined by a similar proportion, the reduction in prosecutions may instead reflect a change in caseload rather than waning enforcement. It is also plausible that external factors—such as the Covid-19 lockdowns and associated restrictions on policing and court operations—contributed to downward fluctuations in numbers during this period.

Nonetheless, when breaches do lead to convictions, sentencing patterns have shifted towards greater use of custody. As shown in Table 1, the proportion of individuals sentenced for breach who received immediate imprisonment increased from about 39% to 52%, from 2017 to 2024, respectively, while the average custodial term rose from roughly four to six months. This trend toward more frequent and longer prison sentences may reflect a judicial shift in treating breaches more seriously once cases reach sentencing—possibly due to greater recognition of the risks to victims from non-compliance, or an increase in aggravated or repeated breaches. Alternatively, it may indicate reduced confidence in, or availability of, community-based disposals for such offences.

However, without information on the nature of the breaches—such as whether they were first-time or repeat failures, involved violence, caused significant harms, or occurred shortly after the order was imposed—it is difficult to identify the factors driving these sentencing changes. Two pieces of evidence suggest that, despite the upward trend, sentencing outcomes may still fall short in addressing the harm caused by breaches. For one, between January 2024 and August 2025, 14 referrals were made to the Attorney General’s Office by the Crown Prosecution Service, victims, families, or members of the public, alleging that sentences for breaches were unduly lenient.6 The challenged sentences ranged from Community Orders to two years’ imprisonment. Although none were overturned (as authorities deemed them to fall outside the statutory remit of the scheme), these referrals indicate that some stakeholders perceive certain sentences as inadequate.

Second, Bates and Hester (2020) caution that prosecuting breaches as standalone offences may, in some cases, obscure a decline in charging for more serious substantive offences—such as assault or stalking—which attract higher penalties. This can occur because proving a breach of a restraining order often involves a lesser evidentiary burden than establishing all the elements of a new substantive offence. While this may make breaches more straightforward to prosecute, it also risks the inadvertent downgrading of threats or harm in criminal justice terms, with potentially less severe penalties imposed, than if the underlying conduct were charged directly. Taken together, the available evidence suggests that breaches of restraining orders are common. While many of these breaches lead to enforcement—and in some cases to increasingly severe sentences—enforcement is inconsistent, public confidence in police action remains limited, and the deterrent effect of restraining orders is far from guaranteed.

Our understanding of the effectiveness of restraining orders is limited by data constraints.

The current evidence base is constrained by:

  • A lack of comprehensive national statistics on the issuance of restraining orders.
  • A lack of proven reoffending statistics when restraining orders are issued or breached, which limits analysis of any potential deterrent effect.
  • No breakdown by the severity or circumstances of breaches, which restricts the ability to distinguish between minor technical breaches and more serious violations that pose substantial risks to victims.

None of the available sources cited herein can provide a complete picture. Figures on police charges, prosecutions, or court outcomes measure only those cases that progress through formal stages of enforcement and are subject to evidentiary and procedural thresholds. They exclude breaches that go unreported, are handled informally, or where victims remain protected without any breach occurring. As such, existing data reflect enforcement activity rather than the overall deterrent effect or protective value of these orders These gaps limit the ability to draw firm conclusions on effectiveness, long-term trends, or enforcement consistency.

Conclusion

Restraining orders issued at sentencing are a vital preventative measure in England and Wales designed to protect individuals from ongoing harassment and threats. Breaches are common and there remains a significant enforcement issue in terms of formal charging or prosecution. Sentencing trends indicate a stronger custodial response over time, but concerns remain about inconsistent enforcement, adequacy of sentencing in certain cases, and the potential for displacement of more serious charges.

Addressing current data gaps is essential for assessing the true impact of these orders. Without regular, transparent reporting, it is difficult for policymakers, practitioners, and the public to evaluate whether restraining orders are meeting their protective purpose, or whether enforcement practices are sufficiently robust to maintain their deterrent effect.

This Bulletin is part of a series on the effectiveness of sentencing in England & Wales. Other Effectiveness Bulletins can be found here.

For further information, contact Annalena Wolcke at a.wolcke@sentencingacademy.org.uk

References

  1. While restraining orders are also available when a defendant is acquitted of a crime (pursuant to Section 5A of the Protection from Harassment Act 1997), this Bulletin concerns only the type issued in connection with a conviction. ↩︎
  2. This statistic was calculated using data in Table 1 of the Bates and Hester (2020) study. ↩︎
  3. These numbers combined restraining orders issued on conviction or acquittal. However, as the response indicates that 97% of all restraining orders issued by this police force during the years 2018-2023 were on conviction, it is most probable that the vast majority of the orders represented in the text also followed convictions. ↩︎
  4. The publication does not mention if these figures include orders issued upon acquittal. ↩︎
  5. The data used in this table were accessed from the data set of the Ministry of Justice’s Criminal Justice System Statistics: Outcomes by Offence data tool for the years 2017-2024, https://assets.publishing.service.gov.uk/media/68878445be2291b14d11affa/Outcomes-by-offence-tool-2017-2024.xlsx. ↩︎
  6. Information collated from a dataset available here: Attorney General’s Office, Outcome of Unduly Lenient Sentence Referrals (2025), https://www.gov.uk/government/publications/outcome-of-unduly-lenient-sentence-referrals. ↩︎