Ten Days: What a Protection Order Is Actually Worth in South Africa

Ten Days

On 12 August 2025, the Temba Magistrate's Court granted Kgaugelo Marota a protection order against her ex-boyfriend, Lazarus Antonio. It barred him from physically abusing her and from entering her home (SABC News, 2025). Ten days later, on 21 August, he found her at a tavern in Maubane village, outside Hammanskraal, and stabbed her to death. He then went home and stabbed his current girlfriend, Ikageng Kgadile, who survived (Sunday Times, TimesLive, 2025). Before leaving the scene of Marota's killing, he filmed her body and posted it online, along with what he called an explanation (The Star, 2025). He was arrested the next day. She was twenty years old.

This is not presented here as an isolated tragedy, though it is one. Rather, it is presented because the order that was supposed to stand between Marota and her killer worked exactly as far as the paper it was written on, and not one step further. Nobody enforced it, since a protection order is not, in itself, an enforcement mechanism. Instead, it is an instruction to the police to enforce something else: to arrest on breach, to remove a firearm, to respond when called. And whether any of that happens depends entirely on a system that South Africa's own Parliament has now spent three years documenting as unreliable, unevenly staffed, and in places simply not functioning.

Two Percent

Start with the harder number first, because it reframes everything that follows. Research from the Remove the Trigger campaign, led by the Saartjie Baartman Centre for Women and Abused Children and drawing on the same 2017 national femicide data South African researchers have used for years, found that only around two percent of women killed by an intimate partner held a protection order at the time they died (Saartjie Baartman Centre, Remove the Trigger campaign, 2025; Abrahams et al, PLOS Medicine, 2024). Marota and Sasha Lee Shah, whose case follows below, are therefore in that rare two percent. Indeed, the overwhelming majority of women killed by a partner never got that far at all: they never applied, never had a hearing, never had a piece of paper with a magistrate's signature on it. The protection order, in other words, is not currently failing at the margins. Rather, it is barely reaching the population it exists for.

That matters because most public conversation about protection orders assumes the opposite problem, namely that they exist and are simply being ignored. That is also true, and it is in fact the part this piece spends the most time on, since it is the part with a clear paper trail and a clear institution to hold accountable. Even so, it sits inside a larger truth that should not get lost: even a perfectly enforced system would still be answering a question for only two women in every hundred. The other ninety eight, meanwhile, are a different policy problem entirely, about access, awareness, cost, and fear of reporting, and therefore deserve their own piece. This one, instead, is about what happens to the two.

Sasha Lee Monique Shah was twenty five years old when she was shot dead by her ex-boyfriend, Kyle Inderlall, in the parking lot of Durban's Gateway Theatre of Shopping on 30 October 2022 (IOL, The Post, 2022). She had been granted an interim protection order against him a little over a month earlier, after telling the court she feared for her life because he owned a firearm and had been stalking her. Inderlall then shot himself. What came out afterward, however, is the part that turns this from a single failure into a pattern. At least three other women or complainants had already gone to police about Inderlall before Shah ever applied for her own order: one report as early as 2020, another woman's protection order application in December 2021 describing threats and an attempted murder charge, and an allegation that he had pointed a licensed firearm at someone (Rising Sun Newspapers, 2025). Furthermore, a Pinetown magistrate ordered that firearm removed when Shah's order was granted, yet it was not removed. Three years later, as a result, the head of the Phoenix Domestic Violence Unit was suspended over the failure, and South Africa's police watchdog, IPID, opened its own investigation into the officer involved (TimesLive, 2025). Shah's mother, meanwhile, is still pursuing that case.

What Non Compliance Actually Looks Like

It would be easy to read the two cases above as unlucky, as individual officers who happened to drop the file. Parliament's own oversight, however, suggests otherwise. In October 2024, the Portfolio Committee on Police, chaired by Ian Cameron, released a briefing covering just six months, October 2023 to March 2024, in which the Civilian Secretariat for Police documented fifty nine separate cases of SAPS non compliance with the Domestic Violence Act (Parliament of South Africa, Portfolio Committee on Police, October 2024). The categories named were specific: failure to complete the register police are required to keep of every domestic violence case, failure to assist a complainant in opening a case, failure to serve a protection order on the person it names, and failure to render what the Act calls a satisfactory standard of service more generally. Moreover, Free State accounted for twenty three of the fifty nine cases, the Western Cape twenty, North West fifteen, and Gauteng one, though a single low count in one province in one reporting window, admittedly, says more about reporting than about safety.

The same briefing found that 97.31 percent of stations visited did have a dedicated interviewing room, which sounds like progress until it is set against the enforcement numbers above it. After all, a room to receive a complaint is not the same thing as a system that acts on what it hears. The Committee's own language on the pattern was, in fact, blunt: it called the non compliance indicative of weak leadership within SAPS, and called for an urgent, station level skills audit (Parliament of South Africa, Portfolio Committee on Police, October 2024). A further detail, moreover, rarely makes it into coverage of this report. Of the officers named in domestic violence matters during that period, 101 were recorded as victims themselves and 191 as perpetrators, which means the institution responsible for enforcing the Act is not, in fact, separate from the pattern it is meant to interrupt.

The Gun That Should Have Been Taken

Firearms sit at the centre of both named cases above, and that is not a coincidence. Roughly thirty five percent of femicide cases in South Africa involve a firearm, according to the same Remove the Trigger research (Saartjie Baartman Centre, 2025), and the law already gives courts the power to address this directly: a protection order can include an instruction to remove any firearm the respondent owns or has access to, precisely because a gun in a home where there is already documented fear is a specific, foreseeable risk rather than a general one. The gap, however, is not in what the law allows. Rather, it is in what happens after the order is signed. In Shah's case, for instance, a magistrate's instruction to confiscate a firearm went unactioned for over a month, and then never happened at all (TimesLive, 2025). What is more, there is no national reporting mechanism that tracks how often this specific instruction is carried out against how often it is issued, which means nobody, not Parliament, not SAPS itself, currently knows the true scale of this particular failure. That absence of data is, in itself, a policy problem, since a gap you cannot measure is a gap you cannot be held accountable for closing.

Why Serving the Order Can Be Its Own Danger

There is a structural problem in how protection orders are served that gets less attention than either of the failures above, and it deserves to be named plainly. In practice, the process of getting an interim order formally served still often depends on the survivor's own information about where the respondent can be found, and in some cases her own presence near that process. Researchers working with UN Women, moreover, have argued that this can turn the order into a flashpoint rather than a safeguard (UN Women, 2024): the moment a man learns he has been formally named and restricted is, after all, also the moment his sense of control is most acutely threatened, at least for a certain kind of abuser. Antonio killed Marota ten days after being served (SABC News, 2025). Inderlall, likewise, killed Shah a little over a month after being served (IOL, 2022). Neither case proves causation on its own, of course, but both fit a pattern researchers have flagged for years: the period immediately after service is disproportionately dangerous, and yet the system currently treats the order as the endpoint of the danger rather than, potentially, the start of a new and acute phase of it that needs its own safety plan.

What Would Actually Need to Change

Four things follow directly from what is documented above, and not from a general call to take gender violence more seriously, which nobody in this conversation actually disputes. First, SAPS needs a specific, publicly reported metric for firearm removal on protection orders: how many were ordered, how many were executed, and within what timeframe, published on the same cycle as the Civilian Secretariat's existing compliance briefings, so that the gap identified above stops being invisible. Second, the fifty nine non compliance cases Parliament documented need individual consequence, and not merely a systemic finding. A skills audit is, admittedly, a reasonable start, yet Jessica Shah is still, three years on, the one pursuing a case against the specific officer involved in her daughter's death, rather than that accountability sitting with the institution by default. Third, the period immediately following service needs its own safety protocol, developed together with the researchers already flagging this pattern, rather than treating service as the moment the case moves from urgent to administrative. Finally, given that non compliance was found across multiple provinces over a period of years, a single province's numbers cannot, therefore, be treated as isolated the next time they appear low. Consistency of enforcement, and not merely its existence on the books, is the actual policy target here.

A Final Word

A protection order is, on paper, one of the more direct tools the South African legal system offers a woman who is afraid of a specific man. It requires her to name him, to describe what he has done or threatened, and to ask a court to formally restrict him. That, in itself, is already a difficult thing to do. What this piece has tried to show, however, is that doing it correctly, and having a court agree, does not currently guarantee very much at all. Kgaugelo Marota, after all, did everything the system asked of her. She named him. The court agreed. Ten days later, nonetheless, none of it mattered. A country that wants that piece of paper to mean something therefore has to fund and staff the institution responsible for making it mean something, and that, ultimately, is a resourcing decision, not a new law.

Figures in this analysis are as at 28 September 2026, the day it was published. For the latest figures, see The Numbers.