When the Law Meets the Lathi: Response to Peaceful Protests in India, 2019-2026. A Legal and Statistical Review

INTRODUCTION

The Constitution of India enshrines the right to assemble peacefully as a foundational democratic freedom. However, the years between 2019 and 2026 produces a recurring and well documented pattern: large peaceful demonstrations, say over citizenship law, farm legislations, the failure of NTA, or any other grievances have been met with baton charges(Lathi Charges), water cannons, tear gas, mass detentions and arrests, and even with gunfire in some instances. This article aims to examine that pattern from two angles: One, the legal architecture that is supposed to govern the conduct of the police at protests; and two, the statistical record of custodial and crowd-control violence that has accumulated alongside it. The article also lays out the perspective of the police and the government, since the record is always contested on several points and not all the clashes between the police and the protesters begin as, or stay one-sided.

India does not maintain an official database of “police violence against protesters.” Instead, we find scattered patchwork data such as National Crime Records Bureau (NCRB) statistics on custodial deaths, National Human Rights Commission (NHRC) complaint registers, parliamentary answers, court judgements, and incident-level reporting by rights organisations and journalists. Each source, however, has different definitions and different blind spots, which itself become a part of the story.

PART I: THE LEGAL FRAMEWORK

1. The Constitutional Starting Point

The right to protest in India flows from three overlapping guarantees in Article 19(1) of the Constitution: freedom of speech and expression (19(1)(a)), the right to assemble peaceably and without arms (19(1)(b)), and the right to form associations or unions (19(1)(c)). None of these rights are absolute in nature. Article 19(3) allows the state to impose “reasonable restrictions” on assembly in the interest of the sovereignty and integrity of India, public order, or related concerns. the entire body of protest-related case law is really a decades-long negotiation over what counts as “reasonable”.

Courts have repeatedly affirmed the protest right in principle. In Himat Lal K. Shah v. Commissioner of Police, Ahmedabad (1973), the Supreme Court held that the freedom of assembly is essential to democratic life and that the state may regulate the time, place, and manner of gathering, for example, for traffic management, but cannot use that regulatory power to impose a blanket ban on peaceful meetings in public spaces. In Ramlila Maidan Incident v. Home Secretary, Union of India (2012), a case that arose from the night-time police crackdown on sleeping protesters at Baba Ramdev’s rally, the Court held that peaceful assembly is a fundamental right that cannot be curtailed through arbitrary executive action, and it emphasised that force, when used at all, must be reasonable and proportionate. Mazdoor Kisan Shakti Sangathan v. Union of India (2018) demonstrates protest zones, buffer areas around sensitive buildings, and a ban on weapons at the site, balancing rights of the protesters and interest of residents. In Anita Thakur v. State of Jammu & Kashmir (2016), the Court again described peaceful protest as rooted in India’s democratic tradition, awarding compensation to protesters who have been assaulted by police.

The 2019-2020 Shaheen Bagh sit-in against the citizenship law produced a more restrictive line of reasoning. In the Amit Sahni v. Union of India judgement (2020), the Supreme Court held that while the right to protest is real, public roads and public spaces cannnot be occupied indefinitely, and that authorities are entitled to seek their vacation through the courts rather than force. This case is frequently cited by both the sides of the debate: protest advocates read it as narrowing the practical space for sustained demonstrations, while the state and its support treat it as a reasonable limit rooted in the competing rights of other citizen.

Taken together, the case law establishes a fairly consistent doctrinal test: peaceful assembly is a fundamental right; restriction on it must be for a legitimate purpose under Article 19(3); and even where dispersal is justified, the force used must satisfy necessity, proportionality, and a “minimum force” standard. Where the record below repeatedly departs from that test is precisely in the proportionality and accountability stages.

2. The statutory machinery for dispersing a crowd

The operational law on breaking up an assembly sits in Chapter X of the old Code of Criminal Procedure (CrPC), sections 129-132 now carried over with renumbering into the Bharatiya Nagarik Suraksha Sanhita (BNSS), which replaced the CrPC on 1 July 2024.

• Section 129 allows an Executive Magistrate, or a police officer not below the rank of sub-inspector, to command any unlawful assembly, defined broadly as five or more people likely to disturb public peace- to disperse, and to use civil force to do so if the command is ignored.
• Section 130 permits the use of armed forces to disperse an assembly when civil force is insufficient and public security demands it, on the order of the senior-most Executive Magistrate present.
• Section 131 allows a commissioned or gazetted armed-forces officer to act on his own initiative, without waiting for a magistrate, if communication with one is not possible and the situation is a genuine emergency, but he must defer to the magistrate's directions the moment contact becomes possible.
• Section 132 grants a form of protection from prosecution for acts done under the preceding sections, but that protection requires prior government sanction to prosecute, not blanket immunity, and courts have read it as applying only to acts done in good faith and in proportion to the threat.

There are two things here. First, prohibitory orders under the old Section 144 of CrPC that deals with public gatherings, curfews and new Section 163 of the BNSS are the most common precursor to a crackdown. Authorities frequently tend to declare an area off-limits to assembly before a march arrives, which then converts an otherwise peaceful gathering into a technically “unlawful” one the moment it proceeds, regardless of it conduct. Second, the “minimum force” and “sanction for prosecution” requirements are meant to be checks on abuse, but in practice the sanction requirement has also functions as a shield: prosecutions of individual officers for excessive force at protests are rare, a pattern that shows up clearly in the custodial-death conviction statistics below.

3. Safeguards against custodial abuse

Separately from crowd dispersal, a body of law governs what happens to people once detained. The Supreme Court’s guidelines in D.K. Basu v. State of West Bengal (1996) require, among other things, that arrests be recorded, that the arrestee’s family be informed, and that medical examination occur at the time of arrest and periodically during detention. These safeguard guidelines aimed squarely at custodial torture and unexplained in-custody deaths. Section 41 and 41A of the CrPC (carried forward in the BNSS) restrict arrest for offences carrying less than seven years’ imprisonment unless specific conditions are met, a provision meant to curb the reflexive mass-arrest of protesters. In compensation jurisprudence beginning with Nilabati Behera v. State of Orissa (1993), the Court established that the state can be held liable in public law to pay compensation for custodial death or torture, independent of any criminal prosecution of the officers involved.

The Supreme Court’s 2006 directions in Prakash Singh v. Union of India additionally called for structural police reforms such as fixed tenures for senior officers, a separation between investigation and law-and-order functions, and independent complaints authorities to review allegations of police misconduct. Implementation across states has been uneven, abd the absence of functioning independent complaints authorities in most states is one of the recurring criticisms in the accountability data discussed below.

PART II- THE STATISTICAL RECORD

1. Custodial Deaths

Custodial death statistics are the most systematically tracked proxy for police violence, even though they capture deaths in lock-ups and jails generally rather than protest-specific incidents. The picture that emerges is consistent across sources:

• NCRB recorded 88 custodial deaths across India in 2021, up from 76 in 2020, with Gujarat recording the highest state total for a second consecutive year.
• Over a longer horizon, a Home Ministry submission to Parliament put the count at 146 police-custody deaths in 2017–18, 136 in 2018–19, and 175 in 2021–22, with Gujarat, Maharashtra, and Uttar Pradesh among the highest state totals over 2017–2022.
• Parliamentary data tabled in March 2026 recorded 170 custodial deaths in India for 2025–26 up to mid-March, with annual totals over the preceding five years ranging between roughly 140 and 176.
• Accountability, by contrast, is strikingly thin. Over a 20-year NCRB dataset spanning 2001–2020, 1,888 custodial deaths were recorded, with cases registered against 893 police personnel, 358 charge-sheeted, and only 26 ultimately convicted. The 2026 parliamentary data was similarly stark: only one disciplinary action was recorded in connection with custodial deaths across the preceding five years nationally.
• There is also a documented gap between what NHRC registers and what NCRB reports from the states, since the NCRB depends on state governments to self-report, and a comparison for 2019 found NCRB recorded 85 police-custody deaths while NHRC's own registry recorded 117 for the same year. Independent monitoring by civil society groups has documented even higher figures in specific years than either official source.

None of this data isolates that occurred specifically during protest-related detentions, which is itself worth noting as a data gap: India’s official statistics are not structured to let a researched pull “custodial deaths of arrested protesters” as a distinct category.

2. Use-of-force and mass-arrest patterns specifically at protests

Protest specific figured come mostly from incident reporting rather than a standing government dataset:

• During the 2019–2020 anti-CAA protests, Human Rights Watch documented at least 25 deaths and hundreds of arrests in the weeks after protests began in December 2019, with all recorded deaths occurring in BJP-governed states; 18 in Uttar Pradesh, 5 in Assam, and 2 in Karnataka. A separate assessment from the International Commission of Jurists put the toll higher, at 31 killed, alongside allegations of indiscriminate firing into crowds, and the use of tear gas and water cannons.
• In the farmers' protest of 2020–2021, violence at the Republic Day tractor rally on 26 January 2021 left more than 300 police personnel injured by protester action, while the farmers' side recorded its own toll through the months-long agitation, including deaths attributed to cold, illness, road accidents, and in one case a lynching within the protest camps unrelated to police action. The October 2021 Lakhimpur Kheri incident, where a vehicle from a minister's convoy ran into a group of protesting farmers, killed eight people and became a separate, heavily litigated case.
• In the 2024–2025 round of farmer protests at the Punjab–Haryana border, more than a hundred protesters were reported injured by pellet-gun and tear-gas use in February 2024, with roughly two dozen police also injured; a 22-year-old protester's death during this period was disputed between farm unions, who attributed it to police action, and Haryana Police, who denied it. Later in 2024, further rounds of tear-gassing accompanied renewed attempts to march toward Delhi.

3. Legal weaponisation as a companion pattern

Alongside physical force, data on anti-terror and public-order law usages shows a pattern of prosecutions that rarely result in conviction but that keep people detailed or under prosecution for extended periods. The United States’ State Department’s 2024 human rights report on India noted that of 1,005 arrests registered under the Unlawful Activities(Prevention) Act (UAPA) in 2022 based on the most recent NCRB data, police themselves dismissed 979 charges, about 97 percent of the underlying charges, with only a small number of convictions from earlier years’ cases and thousands of trials still pending. This pattern of high arrest volume, very low conviction rate, and long pre-trial detention has been a recurring feature of cases against protest organisers and activists linked to the CAA and farm-law demonstrations.

PART III: CASE STUDIES, 2019-2026

Anti-CAA protests (December 2019- March 2020). The passage of the Citizenship Amendment Act triggered nationwide demonstrations. two university crackdowns became emblematic: at Jamia Milia Islamia and Jawaharlal Nehru University in Delhi, and at Aligarh Muslim University, police entered the campuses and used force against students, with doctors at AMU stating that police blocked ambulances from reaching injured students on 16 December 2019. The Allahabad High Court later criticised the Uttar Pradesh Police’s conduct, calling for better training in handling such situations, and ordered compensation for injured students, while also noting that no departmental charges had been filed against the officers involved. Communal violence in northeast Delhi in February killed dozens and became entangled with the protest crackdown, as police subsequently arrested several student activists connected to the anti-CAA protests on charges linking them to the riots, a move rights groups characterised as retaliatory.

Farmers’ Protests (2020-2021) Protests against three farm laws produced a year-long encampment at Delhi borders. The movement was largely non-violent for most of its duration. However, the 26 January 2021 tractor rally, when a faction diverted towards the Red Fort, turned the movement confrontational on both sides. The Lakhimpur Kheri incident later that year in which farmers were run over during a protest connected to a Union minister’s son became a flashpoint for demands of accountability that are, as of the time of this writing, still working through the courts.

Farmer’s protest (2024-2025) A renewed MSP-guarantee campaign centred on Punjab’s Shambhu and Khanauri borders saw repeated use of tear gas and, protesters alleged, pellet-type munitions against a march that the organisers described as a peaceful “jatha” walking under two border points between 16 and 23 February 2024; hospital records attributed three deaths to cardiac arrest, while the cause of a fourth death, that of a 21- year old Shubhkaran Singh, remain undetermined at the time of reporting. Farm unions and state authorities gave conflicting accounts of whether police firing attributed to any of the deaths.

2026 Developments : The “Cockroach” Protests (May-August 2026) The most significant 2026 flashpoint grew out of remarks by the Chief Justice of India in mid-May comparing unemployed youth to "cockroaches" and "parasites of society." A satirical youth movement, the Cockroach Janta Party (CJP), formed within days under political strategist Abhijeet Dipke and rapidly gathered a large online following. It found its cause when a string of national exam irregularities like NEET paper leaks and cancelled tests triggered nationwide anger at Union Education Minister Dharmendra Pradhan, and CJP-led protests, joined by several left-wing student organisations, settled in for weeks at Delhi's Jantar Mantar demanding his resignation. The movement's profile rose sharply after veteran environmentalist and education reformer Sonam Wangchuk joined with a hunger strike, during which he reportedly lost significant weight over roughly 18 days. In mid-July, plainclothes police forcibly removed Wangchuk from the protest site and took him to hospital, citing a court order and his deteriorating health; his wife and CJP disputed this as a disguised illegal detention. That removal escalated the movement, and on 20 July, the opening day of Parliament's monsoon session, more than ten thousand protesters marched toward Parliament. Police met the march with tear gas and baton charges; organisers said at least 150 people were injured, and Amnesty International publicly questioned the legality, necessity, and proportionality of the police response. Authorities closed roughly eighteen Delhi Metro stations near the protest site for several consecutive days, and opposition MPs, led by Rahul Gandhi (who was briefly detained at the protest site and then released), disrupted parliamentary proceedings in solidarity. Wangchuk subsequently ended his fast, and Pradhan resigned shortly afterward, though CJP said its wider protests would continue. The episode is a fresh illustration of the same pattern traced through this piece: a peaceful assembly, framed by police and government statements as a public-order risk, met with force whose proportionality is now itself being publicly and legally contested.

PART IV: READING THE PATTERN

A few threads run through both the legal record and the statistics:

• The doctrine is protective; the practice is inconsistent. Courts have, with real consistency, held that peaceful assembly is a fundamental right and that force must be necessary and proportionate. The gap is not in the black-letter law but in enforcement of that law against the state's own agents.
• Accountability is the weakest link. Whether one looks at custodial deaths (26 convictions from 1,888 deaths over 20 years) or at UAPA arrests (a 97-percent charge-dismissal rate), the data shows a system that investigates and prosecutes its own use of force only rarely, regardless of which party governs a given state.
• Data fragmentation is itself a policy problem. No official dataset isolates protest-related police violence as a category; researchers must triangulate NCRB, NHRC, parliamentary answers, court records, and incident reporting, each with its own gaps.
• The record is not one-sided in every instance. Genuine violence by some protesters like arson, stone-pelting, injuries to police personnel has occurred in several of the episodes above (the January 2021 tractor rally and the 2020 CAA-related riots being the clearest examples), and the state's public-order rationale is not fabricated out of nothing. The legal and statistical critique here is about proportionality and accountability, not a claim that every crackdown was unprovoked.

PART V: REMEDIES : WHAT A PERSON CAN ACTUALLY DO?

If a peaceful protester is beaten, detained, or injured by police, the violation is usually twofold: Article 19(1)(b), the assembly itself was lawful, so no dispersal by force was not justified, and Article 21, the right to life and liberty, which the Supreme Court has long read to include a right to be free from custodial violence and to dignified treatment while in detention. But naming the violation is the easy part. Below is what the law actually offers as a route to a remedy, roughly is the order a person would use them.

1. In the moment: preserve the record

Before any formal process, the practical first step is evidence. Photograph or video injuries and the scene, note the time, date, and location, get the badge numbers or names of officers involved if visible, identify witnesses and get their contact details, and get a medical examination and written medical report as soon as possible, ideally at a government hospital These records are important and become central to any later custodial-violence claim. Under the D.K. Basu guidelines, a person taken into custody is entitled to have a relative or friend informed of the arrest and be medically examined at the time of arrest and at intervals during detention; if these safeguards are denied, that denial itself becomes a part of the evidentiary record.

2. The criminal law route: FIR and criminal complaint

A person assaulted by police can set the ordinary criminal process in motion just as against any other person, since there is no immunity for police officers acting outside the lawful bounds of Sections 129–132 (dispersal) discussed in Part I. Relevant offences (cited here by their old IPC numbers with the current Bharatiya Nyaya Sanhita, 2023 (BNS) equivalents, since India moved to the new criminal codes in July 2024) include voluntarily causing hurt or grievous hurt (IPC 323/325 → BNS 115/117), wrongful restraint and wrongful confinement (IPC 341/342 → BNS 126/127), causing hurt to extort a confession- the specific custodial-torture provision (IPC 330/331 → BNS 120), and criminal force or assault by a public servant exceeding the force lawfully justified by their duty. A refusal by the local police station to register an FIR against fellow officers is itself an offence: a public servant who knowingly disobeys a legal duty in this way can be proceeded against under IPC Section 166A, now BNS Section 199.

If a police station refuses to register an FIR, the law provides two specific escalation routes:

• Written complaint to the Superintendent of Police / Commissioner, under Section 154(3) CrPC, now Section 173(4) of the Bharatiya Nagarik Suraksha Sanhita (BNSS). If the information discloses a cognizable offence, the SP must investigate personally or direct a subordinate to.
• Application to a Magistrate under Section 156(3) CrPC, now Section 175(3) BNSS, asking the court to direct the police to register and investigate the case. This has become a fairly standard route where local police are reluctant to act against their own colleagues.

3. Constitutional and public-law remedies
Because the violation is of a fundamental right, a person can bypass the ordinary court hierarchy and go directly to a High Court under Article 226 or the Supreme Court under Article 32 of the Constitution, seeking a writ. The relevant writs here are:

• Habeas corpus, where a person is illegally detained and their whereabouts or the legality of the detention is in question — this is the fastest available remedy and courts typically list such petitions urgently.
• Mandamus, compelling a police authority to perform a legal duty it has failed to perform (e.g., registering an FIR, conducting a medical exam, releasing someone held beyond the permissible period).
• Certiorari, to quash an unlawful order (for instance, a wrongful detention order or an unlawful prohibitory order used as a pretext for dispersal).

Courts have also developed a distinct public-law compensation jurisdiction, separate from any criminal case against individual officers, under which the state itself can be ordered to pay compensation for a violation of Article 21. This line begins with Rudul Sah v. State of Bihar (1983), where the Court awarded compensation directly in a habeas corpus proceeding for someone illegally detained well past his acquittal, and continues through Nilabati Behera v. State of Orissa (1993), which held the state liable to compensate for a custodial death regardless of whether any officer was ever convicted. Anita Thakur v. State of Jammu & Kashmir (2016), discussed above, applied the same logic to a case of police assault on protesters. The practical significance is that a slow or unsuccessful criminal prosecution does not foreclose compensation. The two tracks are legally independent.

4. The National and State Human Rights Commissions

The Protection of Human Rights Act, 1993 created the NHRC and empowers State Human Rights Commissions (SHRCs) to inquire into human rights violations by public servants, including police, either on a complaint or on their own initiative (suo motu). A complaint can be filed online (nhrc.nic.in) or by post, by the victim or by anyone on their behalf, in English, Hindi, or any Eighth Schedule language, with no fee and no lawyer required, and supported by photos, videos, or medical records. The Commission can summon witnesses, call for records, and recommend compensation, prosecution, or disciplinary action against named officers. Its central limitation is that its recommendations are not legally binding. The concerned government is expected, but not compellable, to respond within roughly a month, which is why an NHRC complaint is often pursued in parallel with, rather than instead of, an FIR or writ petition.

5. Police complaints Authorities

Following the Supreme Court's 2006 directions in Prakash Singh v. Union of India, states are required to set up Police Complaints Authorities (PCAs) at the district and state level, typically chaired by a retired judge, specifically to inquire into serious misconduct such as custodial death, grievous injury, rape, or arrest without due cause fall within their remit. A district-level PCA generally handles complaints against officers up to a certain rank, with the state-level authority handling more senior officers. Implementation has been inconsistent across states as some have never constituted a fully independent PCA as directed. It is worth checking whether a functioning one exists in the relevant state before relying on it as the primary avenue.

6. Free legal aid and civil remedies

Under Article 39A of the Constitution and the Legal Services Authorities Act, 1987, a person unable to afford a lawyer is entitled to free legal aid through the National or State Legal Services Authority (NALSA/SLSA), which can assist with drafting complaints, FIRs, writ petitions, and representation. Separately from any criminal or constitutional process, a person can also bring an ordinary civil suit for damages for assault, battery, or false imprisonment against the officers and, vicariously, the state, though this route is slower and used less often given the availability of the more direct public-law compensation route above.

7. A realistic view of what these remedies deliver

It would be misleading to present this as a smooth pipeline. The custodial-death conviction data in Part II, 26 convictions out of 1,888 recorded deaths over 20 years is itself evidence of how rarely these mechanisms produce individual accountability, even when they exist on paper and are used. What they do reliably deliver, based on the case law above, is a real prospect of compensation (through the Article 32/226 route) and a documented record of the incident (through the NHRC and PCA routes), even where a named officer is never convicted. For someone weighing which avenue to pursue, the practical sequence that lawyers and rights groups typically recommend is: preserve evidence immediately, get a medical exam on record, file the FIR (escalating to the SP or a magistrate if refused), and file parallel complaints with the NHRC/SHRC and, where one exists, the state PCA, with a writ petition reserved for urgent situations (ongoing illegal detention) or for pursuing compensation once the immediate emergency has passed.

Author’s note on sources and limitations of the research

This article draws on various Supreme Court Judgements, NCRB and NHRC data as reported through parliamentary answers and independent analyses, the reporting of Human Rights Watch, Amnesty International, and the International Commission of Jurists, and contemporaneous news coverage of specific incidents. Some factual disputes noted above(for instance, the cause of individual deaths during the 2024 farmers’ protest) remain unresolved or contested between official and civil-society accounts, and this article flags that uncertainty rather than resolving it. Readers relying on this piece for legal argument or advocacy should treat it as a starting map and not a substitute for reading the underlying judgements and primary data themselves.