Why Does India's Dowry Prohibition Regime Fail?

II. Reality Check Through Data

The National Crime Records Bureau logged 6,450 dowry deaths and 13,479 cases under the DPA in 2022, alone. Uttar Pradesh stood first with 2,218 deaths , then Bihar 1,057 and Madhya Pradesh 518.3 But these numbers already hide the larger picture, because quite a few dowry deaths get filed as suicides or accidental burns before any proper probe can reframe them. What seems even more shocking is that, 359 dowry death cases were closed administratively for 'lack of evidence' even as the complaints were regarded as genuine by the investigating officers, while only 4,148 of all registered cases actually progressed to formal charge-sheeting.

The resulting numbers paint an even grimmer picture. Conviction rate: the NCRB's Crime in India 2022 report pegs the conviction rate in dowry-related violence cases at a mere 11–17%. Between 2017 and 2022, 35,493 dowry deaths were reported nationally, an average of 20 a day, yet barely one in six cases ended in conviction. Acquittal rate: in 2024, of all dowry death cases disposed of by courts, 2,024 cases ended in acquittal (50.1%) against 1,869 convictions (46.2%), with 150 cases resulting in discharge. Pendency rate: 3,231 cases remain pending at investigation stage in 2022, declining marginally to 2,765 in 2024. For crimes against women broadly, pendency rate of trials stood at 95% in 2021, meaning courts disposed only 5% of these cases. Taken together, these numbers do not just indict stigma or family pressure. They indict an evidence-generation system that routinely fails before it starts, the proof capable of establishing a case, money transfers, communication trails, medical findings, is either never collected, or is collected too late, after the accused has had the freedom that anticipatory bail affords to steer the story.

Parliament inserted Section 8B into the DPA in 1986, and with that, it effectively created the Dowry Prohibition Officer as the enforcement backbone. Yet the National Commission for Women has described the office as “mainly ceremonial.” Section 8B(1) gives states rather unguided discretion over appointments, and there's no mandatory minimum, no audit obligation that is actually required. The consequences are pretty stark; Kerala had working DPOs in only three out of fourteen districts until a 2022 rule amendment corrected this. No state has ever published consolidated data on DPO inspections or prosecutions that were initiated. An enforcement institution whose output is basically unmeasurable has, in practical terms, no real output.

III. Four Structural Gaps That The Courts Identified And Parliament Ignored

a) The definitional gap: the word 'dowry' is defined broadly in the Act but interpreted narrowly by enforcement agencies.

b) The 'soon before' trap: Section 304B's temporal proximity requirement routinely leads to failed prosecutions when timely documentary evidence is missing.

c) The presumption that cannot be triggered: Section 113B's evidentiary presumption never takes effect because the basic evidence requirement is never satisfied.

d) The bail-as-impunity problem: safeguards against Section 498A misuse have unintentionally allowed accused persons to destroy evidence and avoid cooperating.

The first issue is regarding the definitional gap. In S. Gopal Reddy vs State of Andhra Pradesh 1996 INSC 727, the apex court read the definition of 'dowry' under section 2 of DPA as expansive. It included property 'in connection with' marriage, framed as gifts, within its ambit. This was a significant legal development, yet it failed to address the problem at the investigation stage. For example, in Twisha's case, repeated financial transactions during the marriage seemed ordinary and harmless to the police until CBI intervention. The institution's capacity to enforce the Gopal Reddy interpretation at first contact is lacking. The 'soon before' trigger under section 80(2) BNS requires such harassment to occur 'soon before' death, and this is counterproductive. Reason being the infrastructure to enforce the statute lacks guidance on when a transaction intertwined with harassment should be flagged, until it is too late.

In Kans Raj vs State of Punjab (2000) 2000 INSC 251, the court established a proximity requirement for dowry deaths. The ruling required that there be a living and proximate link between the dowry demand and death. Kaliyaperumal vs State of Tamil Nadu 2003 INSC 428 added a clarification, stating that harassment may not be the immediate cause. A material contribution of harassment towards the circumstances surrounding the death is sufficient to trigger a violation. Together, these cases establish a legal framework. Yet what is lacking is the evidence mechanism, not the doctrine. Because non-circumstantial evidence is hard to establish and lacks contemporaneous documents that aren't rebuttable easily, resulting in acquittal rates eventually. Twisha's WhatsApp chats disclosing harassment, as revealed in the media, precisely cross the proximity standard. However, in most of the cases, no such evidence comes to light, hence no actual convictions.

Next is the untriggered presumption. Section 113B of the Evidence Act (now Section 118 BSA) shifts the burden of proof to the accused once the prosecution establishes dowry demand and harassment within 7 years of a woman's death from marriage. Satbir Singh vs State of Haryana 2005 INSC 417 upheld this requirement once the prosecution establishes the basic evidentiary foundation. The problem is that documentary proof of dowry demand is extremely hard to obtain, so the Section 113B burden never shifts from the prosecution. This way, the protection ensured for victims never gets activated without direct proof of dowry demands or transactions related to the death.

The bail as impunity problem. In Arnesh Kumar v State of Bihar 2014 INSC 463, the Supreme Court laid down that an arrest under Section 498A IPC, without prior compliance with Section 41 CrPC, is unlawful, and also that magistrates must apply their judicial mind before deciding on remand of the accused. The whole judgment was a required corrective as well. Section 498A had been used like a cudgel in matrimonial disputes, meant to annoy and trouble not just the husband but the entire matrimonial household, often on the strength of allegations that were never vetted. And the guidelines initially did curb the abuse to some extent. Still, when it comes to real-life dowry cases, the execution seems to bring a troubling, almost backward effect. By creating these procedural gates before any arrest, Arnesh Kumar, inadvertently gave leeway, often for weeks, for the accused to continue remaining free. That liberty, in practice, means access to physical evidence, access to witnesses, and the ability to coordinate narratives.

Why has Parliament allowed these failures to persist for over six decades? The honest answer is legislative inaction. The DPA has been formally amended only twice, by the Dowry Prohibition (Amendment) Act, 1984 and subsequently in 1986, with a third administrative extension in 2019 that merely brought Jammu & Kashmir within its territorial scope. In three rounds of amendment across six decades, Parliament's exclusive preoccupation has been punishment, raising minimum sentences, tightening bail conditions on paper, and inserting Section 304B into the IPC. What it has never quite legislated for is detection. No real mechanism to spot dowry transfers in real time, no mandatory financial disclosure at marriage, no empowered investigative officer with access to transaction data or anything of the kind. Every amendment keeps asking what to do after a victim suffers. Not one has asked how to see the crime coming.Every dowry transaction eventually ends up inside some financial system. Bank transfers, UPI payments, jewellery purchases property registrations, all of it leaves documentary trails behind. The key argument here is straightforward but nuanced. The DPA's enforcement failure is, at its root, an evidence-generation failure, and financial regulation can generate the evidence that investigations currently cannot track down. India already uses this logic elsewhere too.

IV. The Infrastructure Fix: Regulating The Money

Section 12 of the PMLA already makes it a requirement that each 'reporting entity', which is described as including banking companies, financial institutions, payment intermediaries, and even real estate agents, keeps records of every transaction and also sends Suspicious Transaction Reports to FIU-IND, so where a transaction or a sequence of transactions is viewed as suspicious, or likely to be linked to proceeds of crime. In particular, Section 12(1)(a) says the amount, currency denomination, date, and the involved parties for each transaction have to be maintained. These records need to stay for five years, and they must be kept ready for an investigation. The infrastructure, in other words, already exists. It simply has not been directed at dowry. A targeted amendment to the PMLA's Schedule of predicate offences, or a notification under Section 12 read with Rule 3 of the Prevention of Money-Laundering (Maintenance of Records) Rules, 2005, could designate repetitive wealth transfers from the bride's family to the groom's family proximate to a registered marriage as a category of transaction requiring enhanced due diligence and STR filing. The seven separate transfers spanning October 2025 to March 2026 in Twisha's case would, under such a regime, have generated a documented paper trail before her death. This would have effectively aided the prosecution in establishing a case, pulling the 113B (now Section 118 BSA) trigger.

Three targeted interventions would rebuild the enforcement infrastructure. First, the PMLA's STR framework should be extended to capture patterns of repetitive marriage-proximate wealth transfer as described above. Second, Section 8B of the DPA already sets up the office of Dowry Prohibition Officer but gives it nearly no investigative instruments; granting DPOs supervised access to STR data would allow proactive identification of dowry patterns ahead of time, instead of waiting for a complaint that the victim's family may not lodge because they are too frightened or grieving. Third, marriage registration authorities should be required to flag financial declarations that don't quite line up with income disclosures, creating yet another detection point, right when the marriage is being formalised. The objection is that this kind of financial monitoring steps into private household arrangements. But the reply is simple, the DPA already criminalises those arrangements, the state just currently lacks the tools to actually observe them. Intimidation cannot entirely suppress a bank's STR. Household pressure can't pull back a transaction after it's already been flagged. The proof would, for the first time, exist independently of the victim.

The Supreme Court's judgment in In Re: Enforcement of Dowry Prohibition Act 1961, 2005 INSC 243 stands as the strongest evidence that the DPO approach has failed. There was a writ petition filed in 1997, eleven years after Section 8B was introduced, yet it merely asked states to appoint DPOs, frame rules, and set up Advisory Boards. The Court directed Chief DPOs to submit reports on complaints and prosecutions, and asked DGPs to explain why dowry amounts were not treated as proceeds of crime. The fact that these directions were even needed two decades after Section 8B was brought into force is basically the condemnation itself.

The Twisha Sharma case has not really ended in conviction. It may still, yet the reason it got this far was a CBI probe, High Court intervention, national attention. It isn't about “law working” so much as it is about the accused being a retired judge. That sort of outcome feels less like enforcement, but more like an accident. The DPA has operated for sixty-three years on this foundational idea that nobody has examined hard enough, that the victim, or her family, will produce the complaint, and that the complaint will then produce the evidence by itself. This is fundamentally flawed. The model relies on a grieving family, under social pressure, reaching out to a system that has historically failed them. That is a recipe for continued failure. So until the DPA's enforcement logic shifts, from complaint-driven prosecution to something more institutionalised, like evidentiary trails, say through financial surveillance, empowered DPOs with actual investigative powers, and a bail framework that doesn't give the accused a convenient window to unravel the case, conviction rates will stay dismal. Prosecutions will keep collapsing right at the evidentiary threshold, and the Act's promise will stay exactly as it has always been, symbolic. Sixty-three years of amendments, Supreme Court directions, and parliamentary debates have gone in circles. The law does not need another amendment, it needs a fundamental shift in approach.