Mexico: A Right Won in Court, and the Code That Was Never Changed

Two sentences about Mexico are both true, and a reader who holds only one of them has understood nothing. The first: Mexico’s Supreme Court has held, more than once and unanimously, that punishing a woman for ending a pregnancy is unconstitutional, and it has said so in terms more expansive than most legislatures anywhere have been willing to use. The second: in 2025 the country’s public security statistics recorded seven hundred and twenty-one victims of the offence called feminicidio, and another two thousand and seventy-four women killed in intentional homicides not classified that way.

The temptation is to reach for a contradiction — a progressive court above a violent society — and the chamber declines it, because that reading explains nothing and predicts nothing. This Library has already worked through the version of that argument that runs on a constitution, in the South African essay on the clause that came first and the room it cannot reach. Mexico is not a second copy of that finding. It contributes something the South African case cannot, because in South Africa the protection was written by a constituent assembly and in Mexico it was won in litigation, and those two things are not the same kind of object.

That is the question this essay is actually about, and it is a question with a Japanese edge on it. Where does a right live? It can live in a court judgment, in a statute, or in what a clinic actually does on a Tuesday afternoon. These are three different addresses. A woman does not visit all three. She visits one, and which one she visits is decided by her state, her money, her distance from a city and whether the person at the desk decides to pick up a telephone.

On 7 September 2021 the full bench of the Suprema Corte de Justicia de la Nación resolved acción de inconstitucionalidad 148/2017, brought by the federal prosecutor’s office against provisions of the penal code of Coahuila de Zaragoza. The Court upheld the article that merely defines abortion, on the reasoning that striking it would make forced abortion impossible to charge. It invalidated the article that punished a woman for ending her own pregnancy, and by extension the provision suspending the professional licence of anyone who assisted her, and the provisions that treated the permitted grounds as excuses for conduct still classified as criminal. In the same judgment it invalidated the fraction of the code that set a lesser penalty for rape committed inside a marriage or a concubinage.

That last item is worth pausing on, because it tells you what kind of ruling this was. The same judgment that opened the question of who decides about a pregnancy also removed a discount on the penalty for raping one’s wife. The Court was not treating reproductive autonomy as a separate policy area. It was treating it as part of a single settlement about whether a woman’s body is hers inside a marriage.

Two years later, on 6 September 2023, the First Chamber resolved amparo en revisión 267/2023. The complainant was not a woman who had been prosecuted. It was a civil association, Grupo de Información en Reproducción Elegida, which the Court accepted as having a legitimate interest in challenging the federal criminal framework. The Chamber held the federal abortion provisions unconstitutional and ordered the Congress of the Union to repeal them. And in a third ruling of September 2021, acción de inconstitucionalidad 54/2018, the Court struck down the article of the General Health Law that had written conscientious objection into the statute without limits or safeguards, and told Congress to legislate the limits properly.

Three rulings, three different procedural vehicles, one direction of travel. The direction is real. What it obliges is another matter.

Here is the mechanism, and the mechanism is the argument. An acción de inconstitucionalidad is an abstract challenge: when it succeeds it invalidates the norms that were challenged. Those norms belonged to Coahuila. The judgment did not reach into the penal code of Guanajuato or Durango or Tlaxcala, because those codes were not before the Court.

An amparo is narrower still. Article 107, fraction II of the Constitution carries the principle known in Mexican practice as the relativity of judgments, after the nineteenth-century formula of Mariano Otero: the judgment protects the person who brought it, in the particular case, without making a general declaration about the law. Since the 2011 constitutional reform there is a route from repeated amparo rulings to a general declaration of unconstitutionality, but it is a separate procedure with its own thresholds, and it is not what happens automatically when a chamber decides a case.

What the September rulings did produce is binding criteria. Judges across the country are obliged to follow them. Read that sentence slowly, because it contains the whole counter-thesis. A judge is not someone a woman meets when she wants a procedure. A judge is someone she meets after she has been reported, after a file has been opened in a prosecutor’s office, after she has been questioned. The protection is real and it is located at the far end of a process she would have given a great deal never to enter.

And the statute itself does not move on its own. The official text of the Código Penal Federal published by the Chamber of Deputies, as last reformed on 13 March 2026, still contains articles 329 through 334 exactly as before. Article 332 still provides a reduced sentence of six months to a year for a woman who procures her own abortion, on three conditions: that she is not of bad reputation, that she succeeded in concealing the pregnancy, and that the pregnancy was the fruit of an illegitimate union. Absent any of those, one to five years. The Court declared this unconstitutional in September 2023 and notified Congress in January 2024. In September 2025 a federal district judge refused the Chamber of Deputies a further extension and required officials to certify what had happened to the bill.

One more mechanical fact belongs here, and the essay states it as a reading rather than a proven consequence. In October 2025 Mexico reformed its Ley de Amparo, tightening the definition of legitimate interest so that a complainant must show a real and differentiated injury rather than a hypothetical one. Legitimate interest is the door through which a civil association walked in 2023. A right that was won through a particular procedural door is exposed, in a way a statutory right is not, to changes in the width of that door.

Mexico is a federation of thirty-two entities, each with its own penal code, and abortion has always been state criminal law. By mid-2026, according to the tracker maintained by Grupo de Información en Reproducción Elegida, twenty-four entities had decriminalised voluntary abortion at least within an early gestational window. Mexico City was first, in 2007. Oaxaca followed in 2019, then a cluster in 2021 and 2022, a larger cluster in 2024, and several more in 2025.

The list looks like a legislative tide. Read the routes and it is not one. Coahuila’s code fell to the Court. Tlaxcala’s fell on 2 December 2025, when the Court resolved acción de inconstitucionalidad 89/2024 and invalidated the article that criminalised self-induced and consensual abortion, with retroactive effect for people already prosecuted. In July 2025 a federal collegiate court held the equivalent Durango provisions unconstitutional and ordered the state congress to reform them. Some states legislated. Others were told.

And decriminalisation, in most of these entities, does not mean the offence is gone. It means a window: inside twelve or thirteen weeks it is not a crime, outside it still is. The criminal type stays in the code, which means the machinery that runs on it stays too. Between January 2015 and August 2025 the national public security statistics recorded seven thousand five hundred and eleven investigation files opened for abortion. Three entities account for more than half of them, and the one at the top of the list is Mexico City — the first place in the country to decriminalise, where annual files rose over the decade rather than falling.

That figure carries a limit that matters: the statistics do not say who each file was opened against. A file may concern a woman, or a doctor, or a midwife, or someone who accompanied her. What the figure establishes is not how many women were prosecuted but that the apparatus did not stop when the headline changed. And there is a reason the apparatus keeps being fed. The most common way a woman in Mexico enters a criminal file for abortion is that she went to a hospital, and the hospital reported her.

The chamber now has to be careful, because the violence side of this essay is where the numbers are most often misused, including by people arguing for the right things. There are three distinct instruments in play and they measure three distinct objects.

The first is recorded feminicidio: killings that a state prosecutor’s office has decided to open as that specific offence, reported monthly by the executive secretariat of the national public security system. For 2025 that figure was seven hundred and twenty-one victims. The second is intentional homicide of women: killings of women recorded under the general homicide category, which for 2025 was two thousand and seventy-four victims. Those two together are the roughly seven-and-two-thirds women a day that the annual summaries describe, and the distinction between them is not a fact about the killings. It is a fact about classification — about whether a particular prosecutor’s office opened a particular file one way rather than another. The civil observatory that has tracked this longest reports that of women murdered in the country between 2015 and April 2025, around a quarter of cases were investigated as feminicide.

The third instrument is a survey and it measures something else entirely. The national survey on the dynamics of household relations, conducted by the national statistics institute and most recently published for 2021, estimated that of fifty and a half million women aged fifteen and over, seventy point one per cent had experienced at least one situation of violence in their lifetime, with psychological violence the most reported category. The 2026 round was in the field as this was written and its results are not yet published.

Three numbers, three objects: what prosecutors classified, what prosecutors recorded, and what women said when asked. They cannot be added, they cannot be substituted for one another, and a movement between any two of them can be produced by a change in classification practice with nothing happening in the world at all. Anyone writing that violence in Mexico rose or fell owes the reader which instrument they are reading.

The offence itself is precisely drawn. Article 325 of the Federal Penal Code, as reformed in April 2023, defines feminicidio as depriving a woman of life for a reason of gender, and lists eight circumstances in which a gender reason is taken to exist — signs of sexual violence, degrading injuries or mutilation, prior violence by the perpetrator against the victim, a relationship of kinship or affection or work or trust, threats or harassment, the victim having been held incommunicado, the body exposed in a public place, or the victim having been forced into labour or exploitation. The penalty is forty to sixty years. In April 2026 the Senate approved a constitutional reform empowering Congress to legislate a single national definition, with a proposed range of forty to seventy years, and the anthropologist who drove the original concept into Mexican law publicly questioned parts of the redefinition.

If there is one thing this essay wants a reader in Japan to take from the Mexican vocabulary, it is not the word femicide. It is what the Mexican statute puts inside it. Article 21 of the General Law on Women’s Access to a Life Free of Violence defines violencia feminicida as the extreme form of gender violence, product of the violation of women’s human rights and of the abusive exercise of power, in public and private spheres, which can entail social and state impunity. Impunity is written into the definition. The concept as it entered Mexican law was not a description of bad men. It was a description of bad men plus a state that does not investigate, and the second half was deliberate.

That is the reason this essay does not use machismo as an explanation, and the reason it would be lazy to. An explanation that says a country is violent because of its culture cannot tell you why a killing is classified one way in Sinaloa and another way in Guanajuato, cannot tell you why the same code produces different outcomes in two adjacent municipalities, and cannot be acted on by anyone. The Mexican legal concept was built specifically to point at the part that can be acted on. Whether machismo is a useful category at all is an argument being had inside Mexico, by Mexican women, and a reader here who imports the word without the argument has imported a stereotype.

The same law, in articles 22 to 25, creates the gender violence alert: a set of coordinated emergency measures declared for a defined territory to confront femicidal violence, including measures on prevention, security, prosecution and reparation, with progress reports and a budget. Twenty-five alerts had been declared across twenty-two states by the time the national commission last summarised them, and its own evaluation of the measures found a small fraction fully complied with, a large fraction not complied with at all, and the rest in progress. An independent civil analysis of the alerts found no change in the trends they were declared against, and no sustained improvement in the local capacity — prosecutors’ offices, police stations, personnel — that the measures were supposed to build. Naming an emergency and responding to one are different acts, and Mexico has done the first thoroughly.

The second act has largely been done by women organising. The collective in Veracruz that called the national strike of 9 March 2020, after two killings the country could not look away from, and the Chamber of Deputies that later named the date. The accompaniment networks, of which the oldest and best known was founded in Guanajuato — a state that was among the last to decriminalise — and which have spent a quarter of a century walking women through a medication abortion in places where the law said no, and which after 2022 began doing the same for women in the United States. The green scarves borrowed from Argentina. The song a Mexican songwriter wrote for the marches, which is now sung at them. None of this is endorsed here; it is named, because an essay that describes a country’s violence without naming the people contesting it has written a caricature.

Now the argument against this essay, put as well as it can be put, because it is genuinely strong.

The case for the judicial route is this. A right that depends on a legislative majority disappears when the majority does. A right grounded in a constitutional reading does not, and cannot be traded away in a coalition negotiation. Mexican women obtained in 2021 a holding that no legislature in the country was going to pass, and they obtained it against the political weather rather than with it. Courts exist precisely so that a person whose rights depend on being popular does not have to wait to become popular. The chamber has already found, in the South African essay, that a protection drafted ahead of majority opinion is stronger for not depending on the weather of opinion. That finding applies here too.

And the essay concedes something further. Judicial wins in Mexico have not been symbolic. The 2023 federal ruling changed what a federal health institution can say to a woman at the counter, because the criminal provision it had been citing has been held unconstitutional, and the federal system records tens of thousands of procedures a year. The Tlaxcala judgment reached backwards to people already prosecuted. These are not paper victories.

What the essay holds against the strongest version is narrower and, it thinks, survives. A judicially won right rests on different ground in three specific ways. It requires a litigant, which means it requires an organisation with lawyers, which means the sequence of which states got there first follows the map of where the organisations are. It arrives as an obligation on judges rather than as an instruction to clerks, hospital administrators and prosecutors, who are the people a woman meets. And it can be narrowed by changing the procedure rather than by repealing the right, which is a quieter operation than a vote and attracts less attention.

The headline claim — that abortion is legal in Mexico — is therefore false in a specific way rather than a vague one. It is false for a woman in one of the eight entities that have not decriminalised. It is false for a woman past the gestational window anywhere. It is false for a woman in a town whose only hospital has a director who objects, in a country where the statutory framework for conscientious objection was struck down in 2021 and has not been rewritten. And it is closest to true for a woman in a large city with money, information and a friend who knows which clinic. That is not a country with a right. That is a country in which a right is unevenly reachable, which is a different sentence and the one this essay is willing to sign.

Turn the lens. Japan’s reproductive position is usually described here as settled, which it is not; it is structured, and the structure is a near-mirror of the Mexican one with the layers in a different order.

The criminal articles are in force. Chapter 29 of the Penal Code, articles 212 to 216, makes abortion a crime: a pregnant woman who ends her own pregnancy is liable to up to one year, a person who does it with her consent to up to two years, and a doctor, midwife, pharmacist or medicine seller who does so to between three months and five years. These provisions have been in the code since 1907 and have never been repealed. What sits on top of them is Article 14 of the Maternal Health Act, which permits a designated physician to perform an abortion on two grounds — that continuing the pregnancy would gravely harm the woman’s health for physical or economic reasons, or that she conceived through violence or threat — with the consent of the woman and her spouse. The second paragraph dispenses with the spouse’s consent where he is unknown, unable to express his intention, or has died since the conception. Administrative guidance has since added that consent is not required where a marriage has substantially broken down, as in cases of domestic violence, and that the consent of an assailant is never required; an unmarried woman needs only her own. The statutory text itself has not changed. In its 2024 concluding observations on Japan, the United Nations committee on discrimination against women placed the removal of the spousal consent requirement among the small number of items on which it asked Japan to report back within two years.

Then there is the third layer, and Japan’s is unusually easy to see. The medication regimen of mifepristone and misoprostol was approved in April 2023. Under the notice governing its proper use, as revised in November 2024, it is to be administered in a facility with beds — a hospital or a bedded clinic — for the time being, until the training and safety arrangements are complete; a proposal in 2024 to extend it to clinics without beds was sent back. A woman may be permitted to go home after the second drug only if she wants to, lives within a radius of sixteen kilometres of the facility and inside the same secondary medical district, and returns within about a week. Emergency contraception moved the other way in the same period: approved for over-the-counter switch in October 2025 and on sale from 2 February 2026 without a prescription, with no age limit, but as a category requiring pharmacist guidance — sold only by a trained pharmacist in a pharmacy with a private consultation space and a link to a nearby obstetric clinic, only to the woman who will use it, and taken in front of the pharmacist before she leaves.

Set the two countries side by side and the shape is the same puzzle rotated. In Mexico the permission is in the court and the prohibition is in the statute. In Japan the prohibition is in the statute and the permission is in a different statute, conditioned on a signature from a husband. In both, the layer a woman actually meets is the third one: a designated physician, a bedded facility, a radius measured in kilometres, a pharmacist watching her swallow, a hospital administrator deciding whether to telephone a prosecutor. Neither country’s headline describes that layer. Mexico’s headline says a right was won. Japan’s says the pill was approved. Both are true, and neither is an address.

This chamber has made a related finding before, from the opposite direction, in its essay on the vacuum where a law should be: that where Japan has no statute at all on a question of reproduction, what governs is a professional body’s own rules and a court judgment that asked for legislation and did not get it. Mexico now supplies the companion case. A law that exists and is unenforceable, a ruling that exists and is unimplemented, and a vacuum where nothing was written, all deliver a woman to the same place — a room where somebody with discretion decides.

Against its own interest first. This house sells evenings in which a woman’s stated limit is written down before anything begins and is binding on the person providing the evening, who has no standing to reinterpret it. That product has nothing whatever to do with the layer this essay has just spent eight sections saying is the layer that matters. No arrangement made in this house reaches a hospital administrator, a prosecutor’s office, a pharmacy counter or a penal code. Nothing here should be read as a service that helps with any of it, because it is not one.

Worse than irrelevant, the house has a mild interest in the opposite of this essay’s finding. It sells the proposition that a written rule holds, and this essay has just spent several thousand words on how much has to be true underneath a written rule before it holds for anybody. A reader who finishes this piece slightly less confident in written rules is a slightly harder customer, and the house would rather say that than pretend the tension is not there.

What it does take, and states as an obligation on itself rather than as a selling point, is the address test. Any rule this house writes down can be read for where it actually lives. A rule that lives in a document nobody at the point of contact has read is not a rule. A rule that requires the person it protects to escalate, to explain herself to a third party, or to be believed after the fact, is a rule located at the far end of a process. The Mexican case is the strongest available demonstration that the distance between a written protection and the room where it would have to work is where everything happens, and a house that sells written protections has no business pretending otherwise.

This essay does not claim that Mexico is a violent country, or that violence there has a cultural cause. It refuses that reading explicitly, and it has set out the Mexican statutory concept — which names impunity, including the state’s, as part of what femicidal violence is — as a better analytic instrument than any appeal to national temperament. The argument about machismo is one being conducted inside Mexico by Mexican women, and nothing here settles it or should be read as a verdict on it.

It does not claim that the judicial route was a mistake. It has stated the case for that route at what it believes is full strength, and conceded that the rulings have had non-symbolic effects. Its claim is narrower: that a judicially won right requires a litigant, arrives as an obligation on judges rather than on the people at the counter, and can be narrowed by procedure. Whether the same rights could have been obtained legislatively in the same period is not knowable and is not asserted.

It does not claim a causal relationship in either direction between the legal advance on abortion and the level of lethal violence against women. The two are held in one frame because Mexico holds them in one frame, not because one explains the other.

It does not claim precision it does not have. The count of decriminalised entities moves, and the figure given here is as of mid-2026 from a single tracker. The investigation-file figure does not say who each file was opened against. The three violence measures are described by what they are rather than compared, and the 2026 survey round was still in the field. The Japanese position is stated from the statutory text and the governing administrative notices as they stood in September 2026; guidance changes faster than statute, and a reader should check the current notice rather than this sentence.

It does not endorse any named organisation, movement, official, author or song. Where people and groups are named it is because an account of a country’s violence that leaves out the people contesting it is a caricature, not because this Library has a position on any of them.

And it does not diagnose the reader. Nothing here describes what any particular woman in Japan wants, has experienced, or should do about it. It describes three addresses at which a right can be kept, and observes that the one in the headline is rarely the one at which anybody lives.