Section 498A is not the problem the Court thinks it is

In the decade since Arnesh Kumar², Indian arrest law has been substantially rewritten on the back of a single perception: that Section 498A of the Indian Penal Code is misused. The Supreme Court has now codified the no-automatic-arrest rule for offences punishable with up to seven years' imprisonment, layered on Section 41A notice requirements, run an unsuccessful experiment with Family Welfare Committees in Rajesh Sharma, and produced a substantial corpus of High Court decisions that quash Section 498A FIRs at the threshold. The trajectory looks like an institutional correction to a settled excess. I want to argue that it is not. The misuse claim is real but smaller than its political weight; the data on which the Court has acted has never been seriously tested; the procedural answer the Court has built has cost the women the section was designed to protect much more than it has cost the families the section was designed to deter; and the right answer to the genuine problem is not procedural retreat. It is doctrinal clarity about what the section is for and structural enforcement of the offence the section names.

What Section 498A actually says

It is worth starting with the text, because the Court's misuse jurisprudence has detached itself from the text in a way that has cost the literature precision. Section 498A reads: 'Whoever, being the husband or the relative of the husband of a woman, subjects such woman to cruelty shall be punished with imprisonment for a term which may extend to three years and shall also be liable to fine.'¹ The Explanation defines cruelty in two limbs. The first limb covers 'any wilful conduct which is of such a nature as is likely to drive the woman to commit suicide or to cause grave injury or danger to life, limb or health (whether mental or physical) of the woman'. The second limb covers harassment with a view to coercing a dowry-style demand.

Three features of this text are worth holding in mind through what follows. First, the section is not a dowry-only provision. The first limb covers cruelty whose function is not coercive harassment for property — including, on a long line of Supreme Court authority, mental cruelty, persistent humiliation, deprivation, false attacks on chastity, and the spectrum of patterns recognised in Samar Ghosh.⁸ Second, the section is not a property offence. It is an offence against the body and mental integrity of a married woman in her matrimonial home, where evidence is, by statutory acknowledgement, hard to gather because it is generated behind closed doors. Third, the offence is punishable with up to three years' imprisonment, which places it squarely inside the seven-year band that Arnesh Kumar identified as warranting heightened pre-arrest scrutiny. The Court chose this offence as the test case for that doctrine. That choice was not neutral.

The standard objection to Section 498A is that the second limb of the Explanation reads in dowry-harassment terms and that the section has become a dowry offence in operation. The objection is half right. The section is operationally used in dowry-adjacent cases more often than in pure mental-cruelty cases because the dowry-adjacent allegations are easier to articulate and to corroborate. But this is a feature of police and prosecutorial practice, not a feature of the section, and the case law has repeatedly held that the first limb of cruelty stands on its own. The constitutional challenge in Sushil Kumar Sharma v Union of India was rejected on precisely this reading: the mere possibility of abuse, the Court held, does not invalidate a provision; 'action' may be vulnerable, the 'section' is not.⁴ Pinakin Mahipatray Rawal records the point in terms.¹¹ Reema Agarwal records the related point that the protection extends to women in marital arrangements whose legal validity is disputed.¹² The Section 498A the cases describe is broader than the Section 498A the misuse literature describes, and the misuse literature has shaped the doctrinal response.

The strongest version of the misuse claim

Let me state the misuse case in its strongest form. The case has four propositions and each of them is, taken alone, true.

The first proposition is empirical. The conviction rate under Section 498A, as reported in the National Crime Records Bureau's annual Crime in India volumes, has hovered in the mid-teens for many years.⁷ That is significantly lower than for most cognisable offences. A high acquittal rate is consistent with a substantial proportion of complaints that, on the trial evidence, fail to make out the elements of cruelty within the meaning of the section. If a substantial proportion of complaints are failing at trial, the literature argues, a substantial proportion are failing because they were weak — exaggerated, vindictive, or motivated by considerations external to the marital cruelty the section was meant to address.

The second proposition is anecdotal but consistent. The Court itself has documented over-implication in successive judgments. Preeti Gupta said in terms that 'exaggerated versions of the incident are reflected in a large number of complaints' and that 'the tendency of over implication is also reflected in a very large number of cases'.⁵ The standard pattern the Court has described is the FIR that names the husband, the parents-in-law, the unmarried sister-in-law living three states away, the elderly grandmother who has had no functional contact with the household for years, and one or two distant relatives whose presence in the FIR cannot be plausibly connected to the alleged conduct. The Court's reading of this pattern is that the FIR is being used as an artillery shell — fired in the hope of forcing settlement, capable of catching the actual abusers and a great deal more.

The third proposition is institutional. Once an FIR under Section 498A is registered, the consequences for the named relatives are immediate and serious: arrest, custodial interrogation, the necessity of bail, social stigma, professional consequences. The harm done by an exaggerated or false FIR is therefore front-loaded, and the eventual acquittal does not undo it. The Court has called this 'legal terrorism' in a much-criticised phrase, but the phrase captures, even if hyperbolically, the asymmetry of consequences the FIR imposes.

The fourth proposition is structural. Section 498A is non-compoundable as a matter of central law (with a limited Andhra Pradesh exception), which means that even where the parties reach a settlement, the criminal process continues. The Court has noted this repeatedly and the Law Commission's 243rd Report recommended that the offence be made compoundable with judicial permission and a cooling-off period.⁶ The recommendation has not been implemented.

Taken together, these four propositions produce the strongest version of the misuse claim. There is an offence with a low conviction rate, a known pattern of over-implication, severe front-loaded consequences, and no compoundability route to resolve the dispute short of trial. Something has to give. The Court chose to give on the consequences: tighten the pre-arrest procedure, raise the threshold for FIR registration, and route the complaints through mediation. The route was wrong. The diagnosis was incomplete.

What the Court did not test

The misuse claim is decided by data the Court has accepted on faith. Three holes in the data are large enough that the doctrinal conclusions ought to be reconsidered.

The first hole is the conviction-rate inference. A conviction rate of 15% under Section 498A does not mean that 85% of complaints are false. The 85% figure includes acquittals attributable to compromise after the chargesheet, hostile witnesses (often the complainant's own family members who have come under pressure to retract), prosecutorial failure to lead the available evidence, investigative failure to gather corroborative documentation in time, and the structural difficulty of proving mental cruelty by direct evidence. The Court has noted in Samar Ghosh and elsewhere that mental cruelty is 'difficult to establish by direct evidence' and 'necessarily a matter of inference to be drawn from the facts and circumstances'.⁸ An offence whose elements are inherently difficult to prove at the standard of beyond reasonable doubt will produce a high acquittal rate even where the underlying conduct is widespread and recognised. The inference from low conviction rate to high falsity rate is the inference the Court drew. The data does not support it.

The second hole is in the over-implication evidence. The Court has documented over-implication anecdotally — through quashing petitions and through observations in judgments. It has not documented the rate at which over-implication occurs, the proportion of FIRs in which over-implication is substantive (as opposed to nominal — a relative named in the FIR but never arrested, charged, or subjected to trial), or the proportion of cases in which the over-implication translated into actual custodial detention. The cases that come before the Court are, by selection, the ones in which over-implication has caused harm severe enough to fund quashing litigation. The cases in which the over-implication was nominal — and rapidly cured at the chargesheet stage — do not appear. The data on over-implication is therefore systematically skewed toward the egregious. To build a national arrest rule on that selection is statistically dubious.

The third hole is the counter-factual. The Court has not asked, and on the record before it could not have asked, what the rate of genuine cruelty under-reported because of the procedural rules Arnesh Kumar and Rajesh Sharma introduced. The introduction of the FWC layer in Rajesh Sharma — withdrawn in Social Action Forum for Manav Adhikar in 2018³ — required every Section 498A complaint to be referred to a committee of district worthies before any arrest could be made. The procedure created a substantial barrier between the woman who had filed the complaint and any State action on her complaint. Anecdotal reports from the women's rights bar suggest that the FWC stage was used to pressure complainants into withdrawal, sometimes in the presence of the very relatives who were the subjects of the complaint. The Court did not commission any empirical inquiry into the operation of its own guidelines before retiring them. The harm produced by the guidelines therefore exists in the literature as an estimate, not a finding.

None of this means that the Court was wrong to be concerned about misuse. It means that the data on which the Court built its procedural response was weaker than the response required, and that the cost-benefit calculation the Court implicitly performed was performed on inputs that ought to have been more carefully gathered.

What Section 498A is for

The deeper failure of the misuse jurisprudence is doctrinal. It has lost sight of what Section 498A is doing in the criminal code. The section was inserted in 1983 against the backdrop of a documented epidemic of dowry deaths and a recognition that the ordinary offences against the body — assault, hurt, grievous hurt, abetment of suicide — were insufficient to capture a pattern of conduct that did not always produce visible injury and rarely produced witnesses who would testify against members of the woman's matrimonial household. The section's structural innovation was to recognise that the woman's matrimonial home is a unique evidentiary environment: the conduct happens behind closed doors; the witnesses are family members of the perpetrator; the victim's own family is often unwilling to support a prosecution that will end the marriage. The section's three-year imprisonment ceiling and the inclusion of 'relatives of the husband' in the offender category are designed responses to these features.

The misuse jurisprudence treats Section 498A as if its function were to make arrest available in property disputes between in-laws. That is not its function. Its function is to provide a criminal-procedural backstop against a pattern of conduct that — in the experience of women's rights litigation across multiple jurisdictions — does not stop on its own and does not respond to civil remedies alone. The civil remedies are real and growing. The Protection of Women from Domestic Violence Act 2005 has built out an elaborate machinery of protection orders, residence orders, and monetary relief. The Hindu Marriage Act provides for divorce on cruelty grounds. But the civil machinery operates at a different tempo and threshold from the criminal one, and the women's-rights argument for retaining the criminal route — articulated, defended, and re-articulated through three decades of litigation — is that a regime of civil remedies alone leaves a category of women without recourse whose injuries are precisely those that Section 498A names.

The Samar Ghosh catalogue is the doctrinal evidence for what is meant. The catalogue includes the unilateral decision not to have children, persistent humiliation, refusal of cohabitation, public denigration, false attacks on chastity, and the kind of sustained behavioural pattern that 'creates such anguish that it endangers the life, physical health, or mental health' of the spouse.⁸ The case was decided in the matrimonial-divorce frame, not the criminal frame, but the descriptive content of cruelty it identified is the same descriptive content that Section 498A's Explanation captures. A criminal regime that takes the descriptive content of cruelty seriously will produce an arrest practice that, in cases where the conduct is corroborated to the standards of CrPC Section 41, treats arrest as available. The misuse jurisprudence has produced an arrest practice that treats Section 498A arrest as suspect by default and shifts the procedural burden onto the complainant. The shift is doctrinally backwards.

The cost of getting it backwards

The cost of the procedural shift is borne by the women the section was designed to protect, in three ways.

First, the deterrent effect of the section is calibrated to the credibility of arrest. An offence that does not produce arrest is an offence whose costs to the offender are low, because the social and professional consequences of arrest are the principal price the offender pays. Once Arnesh Kumar and the Section 41A notice rule are in operation, the credible threat of arrest in a Section 498A case is substantially attenuated. The complainant negotiates from a weaker position. The relatives named in the FIR understand that, absent unusual aggravating features, the worst they face is a notice to appear and the prospect of bail at first hearing. The bargaining table tilts.

Second, the Rajesh Sharma Family Welfare Committee experiment — even after its formal withdrawal — left a residue of district-level practice in which 498A complaints are routed through reconciliation processes before any State action is taken. The reconciliation processes are presided over by lay members whose competence to assess matrimonial cruelty is variable and whose default disposition is to recommend the continuation of the marriage. The complainant who has reached the police station has, in practice, already exhausted the family-elder reconciliation route. To send her back into a State-sponsored version of that route is to deny her the State's protective response on the grounds that she has not yet tried what she has tried. The withdrawal of the formal FWC requirement in Social Action Forum did not undo this district-level habit.

Third, the quashing jurisdiction has expanded to a point at which Section 498A FIRs are now routinely quashed under Section 482 CrPC at the threshold, often on settlement terms unfavourable to the complainant. BS Joshi v State of Haryana opened the door; subsequent High Court practice has substantially widened it. The complainant who has filed the FIR finds, several months in, that her continued participation in the criminal process is being held against her in the matrimonial proceedings — that the FIR is being framed as her bargaining chip rather than her grievance, and that the quashing court will accept the framing. The aggregate effect is to convert Section 498A from a criminal offence into a civil instrument of leverage in matrimonial settlement, which is exactly the role the misuse literature accuses the section of having played, except that the conversion has been completed by the courts themselves.

What the right answer looks like

The right answer to the genuine problem the Court was responding to has three components, and none of them is the answer the Court gave.

The first is to make Section 498A compoundable with judicial permission and a cooling-off period, as the 243rd Law Commission Report recommended.⁶ Compoundability would address the structural problem that the section currently has no exit ramp once filed. It would convert the post-filing settlement — which now happens informally and against the complainant's interest at the quashing stage — into a regulated process supervised by the Magistrate, with the complainant's consent verified independently of the in-laws. The cooling-off period would prevent the immediate post-arrest withdrawal that is the most common pattern of pressured retraction. Parliament has not legislated this in fourteen years. It should.

The second is to enforce, not innovate, the Section 41 and Section 41A framework. The framework already requires the police to record a reasoned satisfaction before arrest, to consider whether the conditions in Section 41(1)(b) are met, and to issue a Section 41A notice in the alternative. The Joginder Kumar reasoning is older than Arnesh Kumar and operates at a higher level of generality.⁹¹⁰ The Court's specific application of these principles to Section 498A — through a presumption against arrest in this category of offence and an absolute requirement of pre-arrest Magistrate scrutiny — is what produced the asymmetric harm. A regime that applied Section 41 and Section 41A uniformly, without singling out Section 498A for special pre-arrest treatment, would address the over-implication problem without the side-effects. The over-implicated distant relative is a candidate for a Section 41A notice on the same terms as anyone else. The principal abuser is a candidate for arrest on the same terms as anyone else.

The third is to strengthen, not weaken, the evidentiary infrastructure for prosecuting Section 498A. The high acquittal rate is partly a function of investigative failure — the failure to collect contemporaneous medical documentation, the failure to record neighbours' statements while their memories are fresh, the failure to preserve electronic evidence of harassment patterns, the failure to apply the presumption in Section 113A of the Indian Evidence Act / Section 117 of the BSA where it is available. A police force that prosecuted Section 498A competently would produce a conviction rate higher than the present 15% even without any change to the procedural rules. The misuse jurisprudence has put the entire weight of correction on the front-end, where the cost falls on the complainant, when the correction the section needs is on the back-end, where the cost would fall on the State.

The counter-argument, addressed honestly

The strongest reply to all this is the one that takes the misuse claim seriously as an institutional fact and treats the procedural response as the best the Court could have done with the tools it had. The reply runs as follows. The Court does not have legislative power; it cannot make Section 498A compoundable; it cannot rebuild the police infrastructure for cruelty investigation. What it could do was to tighten the procedural conditions under which the section was enforced. Tightening those conditions has produced costs, certainly, but the costs are concentrated on a category of cases — the marginal, the over-implicated, the strategically filed — in which the harm done by under-enforcement is less severe than the harm done by over-enforcement. The complainant in a genuine cruelty case is not significantly worse off, because the procedural rules permit arrest where the underlying conduct meets the Section 41 conditions; the complainant in a strategically filed case is the one principally affected, and she ought to be affected.

This is the honest version of the reply, and I want to engage it on its terms. What is right about it is that the Court's tools are limited. The Court cannot make the section compoundable; it cannot direct the State to invest in cruelty-investigation training; it cannot rebuild the women's rights infrastructure that has been hollowed out at the district level. The Court's instruments are doctrinal and procedural, and the misuse jurisprudence is the instrument the Court had available.

What is wrong with the reply is its empirical assumption. The assumption is that the procedural rules sort genuine complaints from strategic ones reliably enough that the cost falls on the latter. This is not what is happening on the ground. The complainant in a genuine cruelty case who reaches the police station after months or years of escalating conduct is, in many districts, treated as a candidate for FWC-style reconciliation before her FIR is even registered. The complainant in a strategically filed case is, in many districts, equally able to navigate the procedural rules with the help of competent counsel and to extract settlement value from the threat of arrest even where arrest itself has been deferred. The procedural rules do not sort. They impose a uniform tax on all complainants, with the heaviest incidence on the least-resourced. The misuse literature's worry was that the section was being used as leverage by complainants. The misuse jurisprudence has produced a regime in which the section is leverage for whoever can fund the procedural manoeuvring more aggressively. That is not what the corrective was supposed to achieve.

Where this leaves the section

Section 498A is a small section. It is three sentences, six elements, and a maximum of three years' imprisonment. It does not deserve the doctrinal weight it has been made to carry. The cases it was written for — the dowry harassment, the matrimonial violence, the mental cruelty patterns that Samar Ghosh catalogued — are not problems the Court can solve by procedural means. They are problems the legislature has to take seriously, the prosecution has to investigate competently, and the bar has to defend on the merits. The Court's misuse jurisprudence has tried to do the work of the legislature and the prosecution through arrest doctrine. The work has not been done.

The section will continue to be challenged on misuse grounds. The next round of litigation will probably focus on the bail and quashing standards rather than the arrest standards, because those are the live sites of contestation now. Whoever litigates the next round should ask the Court a question it has not yet been asked: what is the empirical basis for the misuse premise, and what is the empirical evidence on the cost of the procedural correction? The honest answer to both questions is that we do not know. The doctrinal weight the misuse premise has carried in Arnesh Kumar, Rajesh Sharma, and the post-Social Action Forum jurisprudence has been carried on faith. Section 498A's place in the criminal code does not deserve to be settled on faith.

The legislative inertia here matters as much as the judicial activism. Parliament has had three opportunities to make Section 498A compoundable — the 2013 Criminal Law (Amendment) Act, the 2018 amendments to the CrPC on bail, and the 2023 Bharatiya Nyaya Sanhita re-numbering exercise. It has taken none of them. The 243rd Law Commission Report has sat unimplemented for thirteen years. The Bharatiya Nyaya Sanhita has re-enacted Section 498A as Section 85, preserving every operational feature of the original and adding nothing on compoundability. The legislative branch's refusal to engage with the structural problem the Law Commission identified is the political backdrop against which the Court has built its procedural correction. The Court's response was wrong, but the legislative absence that produced the institutional vacuum in which the Court acted is the deeper failure. A Parliament that genuinely believed Section 498A required reform would have reformed it. A Parliament that believed the section was working would have left the misuse-correction to ordinary administrative supervision rather than to constitutional benches of the Supreme Court. The current arrangement — legislative silence, judicial procedural correction, district-level reconciliation practice — is the worst of all available worlds for the women the section is meant to protect.

The section is not the problem the Court thinks it is. The Court has built ten years of arrest jurisprudence on a premise it did not test. The premise is half-true; the response is one-eighth of an answer; the residue is borne by the women the section was meant to protect, who are now told by the institutional Court that the most likely consequence of their complaint is reconciliation in front of a committee of strangers. That is not what 1983 enacted. It is not what Samar Ghosh described. It is what we have done with a section we never read carefully enough.

Get the week's biggest decisions, distilled.

A weekly digest for lawyers, students and the legally curious. Every Friday morning. Free.