UN Standards for Offender Treatment: the Nelson Mandela Rules and the Tokyo Rules

The starting point for both instruments is the Standard Minimum Rules for the Treatment of Prisoners, adopted in 1955 at the First UN Congress on the Prevention of Crime and the Treatment of Offenders in Geneva and approved by the Economic and Social Council shortly after.

The 1955 text was the first attempt at a global statement of minimum conditions in custody: separation of categories of prisoners, minimum standards for cells, food, exercise, medical care and discipline, and limits on the use of instruments of restraint.

It was drafted in the aftermath of the Second World War, when the international community was newly willing to say that how a state treats a person it has locked up is a matter of legitimate international concern, not purely domestic discretion.

The 1955 Rules said almost nothing about people who were not in custody. Probation had existed in some common law systems since the nineteenth century, and community-based sentencing options grew through the twentieth, but no UN instrument addressed the safeguards those measures needed until 1990. That gap mattered because a state expanding non-custodial sentencing without any accompanying standards could just as easily create new abuses (unsupervised, indefinite or disproportionate community obligations) as it could relieve overcrowded prisons.

Two separate instruments therefore emerged from one lineage: the Tokyo Rules in 1990 to cover non-custodial measures, and a thorough revision of the original 1955 text in 2015 to modernise prison standards and rename them the Nelson Mandela Rules.

The revision process for the prison rules ran for several years through an intergovernmental expert group and was adopted by the General Assembly without a vote, a common way for the Assembly to signal broad consensus on a non-binding instrument. The Tokyo Rules have not been revised since 1990, though periodic UN congresses on crime prevention have revisited their implementation.

The 2015 renaming honoured Nelson Mandela, who spent 27 years in South African prisons under apartheid before becoming South Africa's first democratically elected president, and South Africa was among the states that pushed for the revision to bear his name as a statement about the humanity owed to every prisoner, not only political ones.

The 1955 text itself had already been amended once before the 2015 overhaul: the Economic and Social Council, through resolution 2076 of 1977, added a rule extending the Standard Minimum Rules to cover people detained without charge, not only convicted prisoners, showing that even the original instrument was treated as a living text open to incremental strengthening rather than a fixed historical artefact.

The choice of two separate instruments also reflects who carries out each set of rules on the ground. Prison rules are implemented by a single institution inside a space the state fully controls, while non-custodial measures depend on courts, probation officers and often outside employers or volunteers, a far more dispersed set of actors. A combined instrument would have had to write standards general enough to fit both settings, risking rules too vague to enforce in either.

The Nelson Mandela Rules, adopted by UN General Assembly resolution 70/175 on 17 December 2015, apply to everyone held in a prison, whether awaiting trial or already sentenced. The 122 rules are organised into rules of general application, covering every kind of detention facility, and a second part with rules applicable to special categories, such as prisoners with mental illness, women prisoners and prisoners under sentence of death.

Several headline provisions distinguish the 2015 revision from the 1955 original. The most cited is the limit on solitary confinement: Rule 43 bans prolonged solitary confinement outright and Rule 44 defines it as confinement for 22 hours or more a day without meaningful human contact for a period in excess of 15 consecutive days.

Rule 45 further bars its use on prisoners with mental or physical disabilities where it would worsen their condition, and bans it entirely for women who are pregnant, breastfeeding or have infant children, and for children.

The Rules also add a full section on health care that treats prisoners' access to a doctor as equivalent in principle to what is available in the community, and they add explicit protections around searches, disciplinary procedure and access to legal advice that the 1955 text left thin.

The Nelson Mandela Rules also set out an obligation for regular inspection: Rule 83 requires both internal inspections by the prison administration and external inspections by a body independent of it, and Rule 84 gives those inspectors the power to make unannounced visits and interview prisoners in private.

This inspection mandate is one of the clearest links between the Rules and the practical mechanics of oversight bodies such as national human rights commissions and, where a state has ratified it, the Optional Protocol to the Convention against Torture, which sets up National Preventive Mechanisms with a similar visiting function.

The 2015 revision also strengthened complaint and grievance procedures. Rules 56 and 57 give a prisoner the right to make a request or complaint to the prison administration and, if unresolved, to an authority outside it, without censorship of the substance and without being placed at risk of reprisal for having complained.

The revised text ties disciplinary procedure to due process more explicitly than the 1955 original did, requiring that a prisoner facing a disciplinary charge be informed of the charge, given time and facilities to prepare a defence, and given access to legal advice in serious cases.

Read together, the health care, discipline, complaint and inspection provisions make the 2015 text function less as a list of physical minimums and more as a due process framework layered on top of those minimums.

The Rules also address instruments of restraint and body searches, both flashpoints for abuse in the pre-2015 text. Chains and irons are prohibited outright, other restraints may be used only as a precaution against escape or self-harm and never as punishment, and invasive searches are permitted only when strictly necessary and conducted in a manner that respects the prisoner's dignity.

The Tokyo Rules, formally the Standard Minimum Rules for Non-custodial Measures, were adopted by UN General Assembly resolution 45/110 on 14 December 1990. They apply at every stage where a non-custodial option could substitute for detention: pre-trial, sentencing and post-sentencing, and they cover measures such as verbal sanctions, conditional discharge, probation, community service orders, suspended or deferred sentences, and referral to attendance centres.

The Tokyo Rules open with a statement of purpose that runs through the rest of the text: non-custodial measures exist to promote greater community involvement in the management of criminal justice, to reduce the use of imprisonment, and to encourage offenders to have a sense of responsibility toward society. From that purpose flow procedural safeguards that the 1955 prison rules had no reason to address.

The Rules require that any non-custodial measure be imposed only by a judicial or other competent independent authority, and that the offender's consent be obtained where the measure carries an obligation with legal consequences if it is not honoured, such as a mandatory community service order. The Rules also bar measures that involve medical or psychological experimentation on the offender, and require that the offender's dignity be respected at every stage of supervision.

A second set of provisions addresses supervision. the Rules set out that the nature and duration of supervision must fit the offence and the offender, that supervision should encourage participation in education, training and employment rather than crowd it out, and that breach of a condition should not automatically trigger a custodial sentence without a review of the reasons for the breach.

The Rules are explicit that failure to comply is not itself a fresh offence, since the underlying purpose of the Tokyo Rules is to shrink reliance on imprisonment, not to create new routes back into it through technical breaches.

Formal probation itself predates the Tokyo Rules by more than a century. The practice traces to John Augustus, a Boston bootmaker who began vouching for defendants and supervising them informally in the 1840s, and England and Wales put probation on a statutory footing with the Probation of Offenders Act 1907, creating salaried probation officers reporting to the courts.

By 1990 dozens of legal systems, common law and civil law alike, ran some form of probation or community service, but with no shared floor for how those systems should treat the people under supervision, which is the gap the Tokyo Rules were built to close.

The Rules also call on states to develop research, planning, policy formulation and evaluation as part of an effective non-custodial system, treating data collection as a safeguard in itself: a state that cannot show how its community sanctions are actually used cannot demonstrate they are being used to divert rather than to widen the net.

A single set of prison rules could not simply be repurposed for probation and community sanctions because the risks the two settings create are different in kind, not just in degree. Prison rules exist mainly to limit the harm a state can do to someone it has physically confined: overcrowding, physical abuse, denial of medical care, arbitrary discipline.

Non-custodial measures create a different hazard, which criminologists call net-widening: once community sentences exist as an option, courts and prosecutors can be tempted to impose them on people who would previously have received no formal sanction at all, such as a caution or a discharge, rather than using them to divert people away from a custodial sentence they would otherwise have received.

The Tokyo Rules address net-widening directly. The Rules state that non-custodial measures should be used in accordance with the principle of minimum intervention, and the commentary accompanying them explains that community sanctions are meant to substitute for imprisonment, not to expand the total reach of the penal system into cases that would otherwise have stayed outside it.

This is why the Rules insist on judicial or independent authorisation, proportionality between the measure and the offence, and a review mechanism before breach escalates into custody: without those safeguards, a probation order or a curfew can become a parallel system of informal punishment with fewer protections than a prison sentence carries.

The policy driver behind all of this was, and remains, prison overcrowding and the social and fiscal cost of high incarceration rates. By the late 1980s, UN crime congresses were treating overreliance on imprisonment as a policy failure in its own right, not simply a resourcing problem, and the Tokyo Rules were drafted explicitly as a tool for diversion: reducing the prison population by giving courts credible, supervised alternatives, while making sure those alternatives did not become punitive in their own right.

This framing also shaped how the Tokyo Rules expect a sentencing authority to choose between available measures. The Rules call for a graduated range of options rather than a single alternative to custody, so that a court is not forced to choose only between prison and one fixed community order.

The practical effect is a hierarchy: courts are expected to treat a fine, a discharge or a warning as the default response to low-level offending, moving to probation or community service only where a genuine supervisory need exists, and reserving custody for cases where non-custodial measures would not meet the aims of sentencing at all.

Volunteers and the wider community also feature in the Tokyo Rules in a way the Mandela Rules have no equivalent for, since a non-custodial sentence often depends on an employer or a local organisation to carry it out. The Rules encourage states to involve volunteers while insisting they stay subject to the same standards of training and accountability a state employee would face, so outsourcing part of the sentence does not mean outsourcing the safeguards.

Neither instrument is a treaty. A General Assembly resolution does not create binding obligations on member states the way the Convention against Torture or the International Covenant on Civil and Political Rights does once a state ratifies it.

Both the Nelson Mandela Rules and the Tokyo Rules sit in the category international lawyers call soft law: standards adopted by consensus or without a vote, carrying real normative weight because states negotiated and endorsed them, but with no treaty body empowered to hear individual complaints against a state for breaching them.

That does not make them inert. Soft law instruments of this kind function mainly through three channels. First, drafting influence: national prison Acts, prison manuals and probation legislation are frequently drafted or revised with explicit reference to the Rules, so the standards migrate into binding domestic law even though the UN text itself never bound anyone.

Second, interpretive use by treaty bodies and regional courts: the UN Human Rights Committee, the UN Subcommittee on Prevention of Torture, and regional courts such as the European Court of Human Rights and the Inter-American Court of Human Rights have all cited the Mandela Rules when interpreting what a binding prohibition on inhuman or degrading treatment actually requires in a prison context, effectively giving the soft-law text legal bite through a binding instrument it is read alongside.

Third, monitoring practice: national prison inspectorates, ombudsman offices and UN Special Rapporteurs use the Rules as the checklist against which they audit conditions, so a finding of non-compliance carries reputational and sometimes litigation consequences even without a treaty violation being alleged.

The Tokyo Rules are cited less often in litigation, in part because the harms they address (net-widening, disproportionate community conditions) are less likely to produce the kind of individual complaint that reaches a court than a solitary confinement case does.

Probation and parole services and criminal justice reform bodies remain the primary users of the Tokyo Rules, drawing on them when designing community sentencing frameworks or when reviewing whether a jurisdiction's use of non-custodial measures still serves a diversionary purpose or has drifted into supplementing custody rather than replacing it.

The soft-law label also explains why oversight is periodic rather than binding. No standing committee can compel a state to report on compliance the way a human rights treaty body can require state reports under a ratified convention. Compliance information instead surfaces through voluntary national reporting to UN crime congresses and through Special Rapporteurs who visit and publish findings on individual countries, a slower and more diffuse form of accountability than a treaty regime provides.

The two instruments share a lineage and a soft-law status but differ in what they protect against. The Nelson Mandela Rules are a floor on conditions inside an institution the state fully controls: they specify minimums for space, food, health care, contact with the outside world and the limits of permissible discipline.

The Tokyo Rules are a floor on process and proportionality for measures that leave the offender in the community: they specify who may impose a measure, how it must be reviewed, and how a breach must be handled, rather than specifying physical conditions, because there are no physical conditions to specify.

Reading them side by side, the Mandela Rules answer the question of what a state must never do to someone in custody, and the Tokyo Rules answer the question of what a state must never do when it chooses not to use custody at all.

Implementation gaps against the Mandela Rules are well documented across very different legal systems. Overcrowding remains the most common failure: a facility built for a given capacity holding several times that number makes minimum space and health care requirements structurally impossible to meet, regardless of the standards on paper.

Independent inspection, required by Rule 83, is frequently the weakest link, since many national systems still rely on internal prison department reviews rather than a body genuinely independent of the administration being inspected. Solitary confinement limits are also unevenly applied: prolonged isolation persists in some systems under labels such as administrative segregation that are not counted against the 15-day cap because they are classified differently on paper.

Gaps against the Tokyo Rules are harder to measure because non-custodial sentencing data is less consistently collected than prison population data, but two patterns recur. Net-widening is common where community sentencing is introduced without diversion targets, so caseloads for probation services grow without a matching fall in the prison population.

And breach practice is often more punitive than the Tokyo Rules anticipate, with technical non-compliance, such as a missed appointment, sometimes triggering revocation to custody without the individualised review the Rules call for. Reformers in jurisdictions expanding community sentencing, including through recent Indian sentencing reform that has introduced community service as a formal punishment, cite the Tokyo Rules as the reference standard for building those safeguards in from the start rather than retrofitting them after net-widening has already occurred.

Both instruments face a similar structural criticism: soft law can be cited selectively, since a state can treat the Rules as authoritative when a finding suits it and dismiss the same source as non-binding when it does not, with no enforcement body to challenge that inconsistency directly.

Advocates for a treaty-based upgrade argue only binding force would close this gap, while others argue treaty status would slow future revisions and narrow the broad consensus that let the current text pass without a vote, a trade-off that has kept both instruments as soft law for now.