Every legal system that retains capital punishment faces the same structural problem. The penalty is irreversible, but the process that imposes it is run by fallible people working with incomplete evidence. How a system responds to that tension reveals a great deal about what it believes justice requires.

The responses vary considerably. The United States has built an elaborate body of constitutional doctrine around the death penalty, much of it judge-made. India filters cases through a 'rarest of rare' standard and mandatory High Court confirmation. China centralises final review in its Supreme People's Court. Japan combines lay-judge trials with long periods of confinement and executions disclosed to prisoners only hours beforehand.

This piece compares three layers of that architecture: the procedures that surround a capital case, the legal tests applied to execution methods, and the categories of people placed beyond the penalty's reach. Each layer shows a different way that law tries to make an absolute punishment compatible with conditional knowledge. It also shows why, in many places, that effort eventually gave way to abolition.

American scholars use the phrase super due process to describe the additional procedure that capital cases attract. The idea grew out of the Supreme Court's response to Furman v. Georgia (1972), which found existing death penalty statutes applied arbitrarily. In Gregg v. Georgia (1976), the Court approved revised schemes featuring bifurcated trials, defined aggravating factors, and appellate review. Woodson v. North Carolina, decided the same day, struck down mandatory death sentences and required individualised consideration of each defendant.

India reached a similar destination by a different route. In Bachan Singh v. State of Punjab (1980), the Supreme Court upheld capital punishment but confined it to the 'rarest of rare' cases, requiring courts to weigh aggravating and mitigating circumstances. Indian procedure also makes appellate scrutiny automatic: a sessions court's death sentence cannot be carried out until a High Court confirms it. That requirement carried over from the old Code of Criminal Procedure into its 2023 successor.

China's approach is more centralised. Since 2007, every death sentence must be reviewed by the Supreme People's Court, reversing an earlier delegation of that power to provincial courts. Chinese law also permits a death sentence with a two-year reprieve, usually commuted to life imprisonment, which lets courts express maximum condemnation while executing far fewer people. Japan, by contrast, has no mandatory appeal, and defendants may withdraw their own appeals.

The common thread is that safeguards tend to accumulate after errors become visible. Japan's 2024 retrial acquittal of Iwao Hakamada, sentenced to death in 1968 and confined for decades, renewed debate there about retrial procedure. Safeguards also bring their own costs. Lengthy appeals produce years on death row, which the Privy Council in Pratt and Morgan (1993) held could itself amount to inhuman treatment. Better process can make the penalty both fairer and harder to carry out.

Takeaway

Every safeguard added to a capital process trades speed for accuracy. A system that takes accuracy seriously enough may find that the penalty itself becomes practically unworkable.

Nearly every constitutional order that retains capital punishment also prohibits cruel punishment. Courts must therefore decide when a method of killing becomes unlawfully cruel. The question is rarely whether death itself is cruel, since most retentionist courts settled that early. Instead, they ask whether the manner of execution adds suffering beyond what the sentence requires. That narrow framing shapes everything that follows, because it treats the problem as one of technique rather than principle.

The United States Supreme Court has set a demanding test. In Baze v. Rees (2008), a plurality upheld Kentucky's three-drug lethal injection protocol, asking whether it posed a substantial risk of serious harm. Glossip v. Gross (2015) added that a challenger must identify a known and available alternative method. Bucklew v. Precythe (2019) stated plainly that the Eighth Amendment does not guarantee a painless death. Courts compare methods rather than judging any single method in isolation.

That comparative test has produced an unexpected consequence. As European export restrictions and pharmaceutical company policies made lethal injection drugs harder to obtain, several states turned to alternatives. Alabama carried out the first execution by nitrogen hypoxia in 2024, and South Carolina conducted a firing squad execution in 2025. A doctrine designed to evaluate methods has instead encouraged legislatures to search for new ones.

Japan and India illustrate a more deferential model. Japan's Supreme Court upheld hanging in 1955, and the procedure is still governed in part by an 1873 proclamation. India's Supreme Court upheld hanging in Deena Dayal v. Union of India (1983), though in 2023 it invited evidence on whether less painful methods exist. In these systems, method challenges rarely succeed, partly because courts treat the legislature's choice as presumptively valid rather than demanding comparative proof.

Takeaway

A cruelty test framed as a comparison between methods can never conclude that execution itself is the problem. It can only ask whether there is a better way to do it.

International law sets a floor. Article 6 of the International Covenant on Civil and Political Rights prohibits death sentences for crimes committed by people under eighteen and bars executing pregnant women. Most retentionist states have adopted these rules, but many add categories of their own. The choice of which groups to exempt reveals each system's underlying theory of moral responsibility, and the reasoning differs more than the outcomes suggest.

The United States has built its exemptions through constitutional litigation measured against 'evolving standards of decency.' Atkins v. Virginia (2002) barred executing people with intellectual disability, and Roper v. Simmons (2005) did the same for offenders under eighteen at the time of the crime. Ford v. Wainwright (1986) prohibited executing someone who cannot understand their punishment, and Kennedy v. Louisiana (2008) excluded non-homicide crimes against individuals. Each rests on reduced culpability or proportionality.

Other systems draw lines by status rather than capacity. Belarus exempts women entirely, alongside men aged sixty-five or older at sentencing. China's 2011 criminal law amendment generally excluded defendants aged seventy-five or older at trial. These age and gender exemptions reflect judgements about vulnerability or social role. They also invite questions about equal treatment that capacity-based exemptions largely avoid, since they distinguish between offenders whose culpability may be identical.

In practice, exemptions are only as strong as the evidence that triggers them. The US Supreme Court had to intervene in Hall v. Florida (2014) and Moore v. Texas (2017) after states applied unduly narrow tests for intellectual disability. In countries where birth registration is incomplete, disputes over a defendant's age at the time of the offence can decide whether the juvenile exemption applies at all. A categorical rule depends on fact-finding that is anything but categorical.

Takeaway

Who a system exempts from execution tells you what it believes makes a person fully responsible, and how much it trusts its own courts to measure that.

Across these jurisdictions, a pattern emerges. Retentionist systems rarely defend capital punishment as unconstrained. They defend it as constrained by procedure, method review, and categorical limits. The legitimacy of the penalty is made to depend on the reliability of those constraints.

That dependence is also a vulnerability. Each safeguard adds delay, each method challenge exposes supply and design problems, and each exemption requires contested fact-finding. Several jurisdictions that abolished capital punishment did so partly because these constraints became too costly or too unreliable to maintain.

For anyone evaluating a legal system, the useful question may not be whether it permits execution. It may be whether its constraints are honest about the limits of what courts can know.