TL;DR: Bail decides one narrow question: whether an accused person waits for trial in custody or out of it. It is not a verdict on guilt. In every non-bailable case the court runs a “triple test”, asking whether you are a flight risk, whether you might tamper with evidence, and whether you might influence witnesses. The gravity of the offence is one factor the court weighs, never the only one, and the Supreme Court has held that gravity alone cannot decide bail. The headline principle is old and still binding: “bail is the rule, jail the exception”, rooted in the personal liberty guarantee of Article 21. None of this changes because a case is on prime-time television. Public sentiment and media glare are, in law, irrelevant to the bail calculation, and the Court has called trial by press the “very antithesis of rule of law”. This piece explains how the law is supposed to work in the cases everyone is watching, with the cases and statutes that fix the rules.
On this page
- What bail actually decides, and what it does not
- Bail, anticipatory bail, and the kinds in between
- Bail is the rule, jail the exception: where it comes from
- The triple test: the only questions a bail court should ask
- Gravity of the offence: a factor, not a trump card
- Why media glare and public sentiment should not move the needle
- When the statute makes bail harder: PMLA, UAPA and the rest
- How a watched case moves through the bail process
- What recent Supreme Court bail orders tell you
- Bail and anticipatory bail side by side
- How Niyam helps you research bail law
- Frequently asked questions
What bail actually decides, and what it does not
Start with the thing most coverage gets wrong. A bail order says nothing about whether the accused did it. It answers a single question: pending the trial, does this person sit in jail or stand outside it? That is the whole of it.
The reason this matters is the presumption of innocence. Until a court convicts you, the law treats you as innocent. Pre-trial custody is therefore not punishment. It is a precaution, justified only when the state shows a real risk that letting you out would wreck the trial. When you read a bail order as a referendum on guilt, you have already misread it.
This is not an abstraction in India. According to the National Crime Records Bureau’s Prison Statistics India 2022, undertrial prisoners made up 75.8 percent of the prison population, that is 4,34,302 people out of 5,73,220, held in jails running at 131 percent of capacity. As LiveLaw reported on the same data, most people behind bars on any given day have not been convicted of anything. Bail is the valve that is supposed to keep that number from being even larger.
So when a famous defendant is refused bail, the question to ask is narrow and technical: did the court find a real risk to the trial, or did it deny liberty for some other reason the law does not recognise? That second possibility is exactly what the Supreme Court keeps warning trial courts against. Hold on to the distinction between custody and conviction. Almost every confusion about high-profile bail flows from collapsing the two.
A second point worth fixing early. Bail jurisdiction is split across courts. A magistrate, a Sessions court, the High Court and the Supreme Court can each be the right forum depending on the offence and the stage, and a refusal at one level is not the end of the road. The accused can climb. That is why a single case can produce a string of bail orders that seem to contradict each other. They are different courts answering the same narrow question at different moments, on different records.
Bail, anticipatory bail, and the kinds in between
“Bail” is a family of remedies, not one thing. Knowing which one is in play tells you what the court is actually deciding.
Regular bail is release after arrest. You are in custody, and you ask the court to let you out pending trial. For non-bailable offences this is governed, under the new criminal code, by Section 480 of the Bharatiya Nagarik Suraksha Sanhita, 2023, which replaced Section 437 of the old Code of Criminal Procedure. The High Court and Sessions court have wider special powers to grant or modify bail under Section 483 BNSS, the successor to Section 439 CrPC.
Anticipatory bail is pre-arrest protection. You apprehend arrest in a non-bailable case and ask a court, before the handcuffs, for an order that you be released on bail if and when arrested. This lives in Section 482 BNSS, which carried forward Section 438 CrPC. Only the High Court and the Court of Session can grant it. The standard conditions are familiar: make yourself available for interrogation, do not threaten or induce any witness, and do not leave India without the court’s permission. The full mechanics are set out in anticipatory bail under BNSS Section 482.
There are narrower species too. Interim bail is a short, temporary release while a fuller application is pending. Default bail, also called statutory bail, is an automatic right that accrues when the investigating agency fails to file its charge sheet within the time the law allows, a remedy explained in default bail under BNSS Section 187. And the whole architecture rests on a prior classification: whether the offence is bailable or non-bailable, where bailable means release is a right and non-bailable means release is a matter of judicial discretion, a line drawn out in bailable versus non-bailable offences.
For the cases that dominate the news, two of these dominate the courtroom. The accused who sees the storm coming files for anticipatory bail. The accused already arrested files for regular bail. Everything else is a variation. If you want the larger reorganisation of criminal procedure that renumbered all of this, see the overview of the new criminal laws: BNS, BNSS and BSA.
Bail is the rule, jail the exception: where it comes from
You have heard the phrase. It is not a slogan invented by defence lawyers. It is settled Supreme Court law, and it traces to Justice V.R. Krishna Iyer in the late 1970s.
In Gudikanti Narasimhulu v. Public Prosecutor, decided in 1978, Justice Krishna Iyer framed bail as a question of “liberty, justice, public safety and burden of the public treasury”, and insisted that personal liberty is too precious to be denied merely because of the nature of the accusation. The companion ruling, State of Rajasthan v. Balchand, put it in the line everyone now quotes: the basic rule is bail, not jail, except where there are circumstances suggestive of fleeing from justice, or thwarting the course of justice, or repeating offences, or intimidating witnesses.
Read that exception clause carefully, because it is the whole game. Krishna Iyer did not say liberty is absolute. He said custody is the exception, and the exception has to be earned by pointing to a concrete danger. The burden sits on the side arguing for jail, not on the accused arguing for freedom. That is the inversion that “bail is the rule” actually encodes.
The principle has a constitutional home. The right to personal liberty under Article 21 of the Constitution is what makes pre-trial detention something the state must justify rather than something it may impose at will. The connection between the bail rule and Article 21 is developed at length in bail is the rule, jail the exception under Article 21.
Two cautions keep this honest. First, the rule is a principle, not a guarantee, and it bends when a statute deliberately raises the bar, which we come to below. Second, courts and commentators have noted a gap between what the principle promises and what trial courts deliver. The Supreme Court itself, granting bail to a former Delhi deputy chief minister in 2024, said trial courts and High Courts had “forgotten” the rule, language we will look at directly. The principle is binding. Its application is uneven. Both things are true at once.
The triple test: the only questions a bail court should ask
Strip away the noise and a bail decision in a non-bailable case runs through three questions. Lawyers call it the triple test, or the tripod test. The Supreme Court applied it squarely in P. Chidambaram v. Directorate of Enforcement in 2019, and trial courts use it every day.
The three questions are simple to state.
Is the accused a flight risk? The court asks whether there is a real chance the person will abscond and dodge the trial. Roots in the community, a fixed address, family, a passport that can be surrendered, a record of appearing when called: all of these cut against flight risk.
Will the accused tamper with evidence? The concern is interference with the investigation, destruction of documents, or doctoring of material the trial depends on. The court looks at whether the evidence is still in the accused’s reach or already secured by the agency.
Will the accused influence witnesses? This asks whether the person can pressure, threaten, or buy off those who will testify. The danger is highest where witnesses are subordinates, employees, or otherwise within the accused’s sphere of power.
That is the entire core of a bail inquiry. Notice what is not on the list. The court is not asked whether the accused is guilty. It is not asked whether the public is angry. It is not asked whether the case is on television. The triple test is a risk assessment about protecting the trial, not a preview of the verdict.
| Question the court asks | What it is testing | What is irrelevant to it |
|---|---|---|
| Flight risk | Will the accused abscond and evade trial? | Whether the public thinks the accused is guilty |
| Evidence tampering | Can the accused destroy or doctor evidence? | The volume of media coverage |
| Witness influence | Can the accused threaten or buy witnesses? | Online petitions or street protests |
| Guilt or innocence | Not asked at the bail stage | Decided only at the trial, on evidence |
When a bail order is well written, you can map its reasoning onto these rows. When you cannot, when the order reads like a moral verdict or a response to outrage, that is usually the sign of an order that an appellate court may set aside. Reading orders this way is a skill, and the method carries over from how to read a judgment.
Gravity of the offence: a factor, not a trump card
Here is where high-profile cases get genuinely difficult, and where a lot of public commentary goes wrong. Does it matter that the alleged crime is serious? Yes. Does seriousness by itself justify keeping someone in jail until trial? No. The Supreme Court has been clear on both halves.
In P. Chidambaram v. Directorate of Enforcement, the Court held that “the gravity of the offence alone cannot be decisive ground to deny” bail, while adding that in an economic offence with a severe impact on society, courts should be slow and careful before granting it. Both sentences live in the same judgment. Gravity raises the level of scrutiny. It does not switch off the triple test.
The cleaner statement comes from the 2G spectrum prosecution. In Sanjay Chandra v. CBI, decided in 2012, the accused faced grave economic charges, and the Supreme Court still granted bail. The Court held that the seriousness of the charge and the severity of possible punishment are relevant, but bail is not to be withheld as a form of advance punishment, and that the object of holding a person before trial is to secure their presence, not to punish them. Pre-trial detention, the Court said, cannot be punitive.
Put the two together and the rule emerges. Gravity is an input into the triple test, not a substitute for it. A serious charge may make flight more tempting, or tampering more consequential, and a court may weigh that. But the court must still find an actual risk. It cannot reason “the crime is terrible, therefore no bail”, because that reasoning convicts before trial and turns custody into a sentence served before any finding of guilt.
This is also why the same offence can produce opposite bail outcomes. Two accused facing identical charges can get different orders because one has a passport and offshore links and the other has a fixed home and a clean record of attendance. The charge is the same. The risk profile is not. A bail court is supposed to be looking at the second thing.
Why media glare and public sentiment should not move the needle
Now the heart of it. In a sensational case, the loudest fact in the room is the coverage. Prime-time anchors run verdicts every night. Hashtags demand jail. None of that is a legal input into bail, and the Supreme Court has said so in strong terms.
In State of Maharashtra v. Rajendra Jawanmal Gandhi, decided in 1997, the Court warned that “a trial by press, electronic media or public agitation is the very antithesis of rule of law” and can lead to a miscarriage of justice. The bench that decided that case was telling judges, in plain words, that the job is to apply law to evidence, not to read the mood of the crowd. A judge who denies bail because the public is furious has substituted sentiment for the triple test.
The Court went further on the mechanics in Sahara India Real Estate Corporation v. SEBI in 2012. There it recognised that an accused or an aggrieved person who genuinely fears that publicity will prejudice a fair trial can ask a constitutional court to postpone reporting of particular phases of a case. The Court held that freedom of expression under Article 19(1)(a) is not absolute and can yield, for a limited time, to the fair-trial guarantee of Article 21. The point is not to gag the press permanently. It is to stop coverage from contaminating the very process the court is running.
There is a deeper logical reason media pressure cannot count. Coverage is not evidence. It is not tested by cross-examination, it is not on oath, and it is frequently wrong. If a bail court could be moved by how a story is being played, then the people with the best public relations would get the best bail outcomes, and liberty would track press cycles instead of risk. That is the opposite of equal justice. The remedy for prejudicial coverage, where it crosses into interfering with proceedings, sits in the law of contempt of court in India, not in the bail order.
So the honest reading of a high-profile bail decision treats the media as background, not as a party. The risk to the trial is the same whether the case is anonymous or trending. A court that lets the trending part change the answer has stopped doing law. That is the standard the cases set, even if it is not always the standard every order meets.
When the statute makes bail harder: PMLA, UAPA and the rest
The default rule is “bail is the rule”. Parliament has carved out exceptions, and many of the cases that fill the headlines fall inside those carve-outs. This is the single most important thing to understand about famous bail fights, because it explains why the same Supreme Court that says “bail is the rule” can also uphold prolonged custody.
Certain special statutes attach “twin conditions” to bail. The Prevention of Money Laundering Act, 2002, is the leading example. Under Section 45 of the PMLA, a court cannot grant bail unless the public prosecutor has had a chance to oppose it, and the court is satisfied that there are reasonable grounds for believing the accused is not guilty and is not likely to commit any offence while on bail. The Supreme Court upheld these conditions in Vijay Madanlal Choudhary v. Union of India in July 2022, reasoning that money laundering threatens the financial system and warrants a stringent regime. Similar bars exist under the Unlawful Activities (Prevention) Act through Section 43-D(5), under the NDPS Act through Section 37, and under a handful of other special laws.
The twin conditions invert the usual posture. Instead of the state showing why you should be jailed, you must persuade the court there are reasonable grounds to believe you are not guilty, at the bail stage, on the material so far. That is a heavy thing to ask of a defence, and it is why bail under these statutes is genuinely harder to get.
But the carve-outs are not absolute, and the Constitution pushes back. In Union of India v. K.A. Najeeb, decided in 2021, the Supreme Court held that the rigours of the UAPA bail bar “melt away” where an accused has suffered prolonged incarceration and there is no realistic prospect of the trial finishing within a reasonable time. The Court held that statutory restrictions do not oust the power of constitutional courts to grant bail when continued detention would violate the Article 21 right to a speedy trial. Delay, in other words, can defeat even a stringent bail bar, because the Constitution sits above the statute.
This is the tension that runs through every famous money-laundering or terror-law bail case. The statute says “prove you are probably innocent”. The Constitution says “but you cannot rot in jail for years without a trial”. Courts hold the two in uneasy balance, and which way a given case tips depends on how long the person has been inside and how far the trial has actually moved.
How a watched case moves through the bail process
It helps to see the sequence, because the news usually shows you only a fragment of it. Take a neutral, real example: the bail litigation of a former Delhi deputy chief minister in the Delhi excise policy matter, which produced a clear Supreme Court ruling and so makes a clean teaching case without anyone needing to relitigate guilt.
The accused was arrested by investigating agencies and held in custody. He applied for bail. The trial court refused. He went to the High Court. The High Court also refused. He reached the Supreme Court, which in Manish Sisodia v. Directorate of Enforcement, reported as 2024 INSC 595 and decided on 9 August 2024, granted bail. That is the normal shape of a serious bail fight: refusal, refusal, then a higher court taking a different view on the same record. The string of “no” orders before the “yes” is not a contradiction. It is the appellate structure doing its work.
The reasoning is what makes the case useful. Justices B.R. Gavai and K.V. Viswanathan noted that the accused had been in custody for around 17 months without the trial even beginning, and held that he had been deprived of his right to a speedy trial. As the Supreme Court Observer recorded the ruling, the bench rejected the agency’s suggestion that the accused should keep shuttling between courts, comparing it to a game of snakes and ladders. The Court said, in words quoted in Verdictum’s report on the judgment (2024 INSC 595), that bail is not to be withheld as a punishment, and that it was “high time” courts recognised that “bail is the rule and jail is the exception”.
Two lessons travel out of this example to every sensational case. First, prolonged pre-trial custody with no trial in sight is itself a ground for bail, independent of the merits, because it converts detention into punishment. Second, the seniority and notoriety of the accused did not change the legal test. The Court ran the same analysis it would run for an unknown undertrial. That is the point of the framework. It is supposed to be indifferent to fame.
The general guidance for how courts should handle bail across categories of offence was systematised in Satender Kumar Antil v. CBI, decided in 2022, where the Supreme Court sorted offences into categories and laid down directions to stop the reflexive denial of bail and the routine arrest of people who had cooperated with the investigation. If you are drafting rather than just reading, the structure of a persuasive application is set out in the guide to drafting a bail application.
What recent Supreme Court bail orders tell you
The last two years have been an unusually rich period for bail jurisprudence, and the pattern is worth seeing whole rather than one viral order at a time.
The Supreme Court Observer, reviewing the Court’s 2024 work, titled its survey batting for bail in the face of delayed trials, and the phrase captures the year. In several high-visibility prosecutions under stringent statutes, the Court granted bail by leaning on the length of incarceration and the slim odds of a quick trial, rather than on the strength or weakness of the allegations. Delay did the work that the merits could not.
That trend is not unbroken, and the data complicate the headline. A separate Supreme Court Observer study assembling a dataset of the Court’s UAPA and PMLA bail decisions across 2024 and 2025 shows real divergence between benches and between statutes, with the principle stated confidently in some orders and applied thinly in others. LiveLaw’s roundup of 25 notable Supreme Court judgments of 2024 on bail and arrest shows the same range: strong liberty-protective rulings sitting beside cautious ones, often within weeks of each other.
For a working lawyer or an informed reader, the takeaway is not “bail has become easy” or “bail has become hard”. It is that the controlling principles are stable while their application is contested, which means the specific judgment you cite matters enormously. The right precedent for your facts may be Najeeb on delay, Sanjay Chandra on economic-offence gravity, or Antil on category-based directions, and picking the wrong one weakens an otherwise good case. To watch this develop month by month, the running summaries in the Supreme Court digest track the latest benches.
One more thread runs through all of it. Before you rely on any bail precedent, confirm it has not been distinguished, doubted, or overtaken. Vijay Madanlal upheld the PMLA twin conditions, but later benches have read them more flexibly, which means a 2022 proposition can be softer in 2026 practice. Checking whether a case is still good law is not optional housekeeping in this area. It is the difference between an argument that holds and one that collapses on the first rejoinder.
Bail and anticipatory bail side by side
The two remedies that dominate famous cases differ in timing, forum, and standard. Holding them next to each other prevents the common mistake of asking the wrong court for the wrong thing.
| Feature | Regular bail (Section 480 / 483 BNSS) | Anticipatory bail (Section 482 BNSS) |
|---|---|---|
| When you seek it | After arrest, while in custody | Before arrest, in anticipation of it |
| Which courts can grant it | Magistrate, Sessions, High Court, Supreme Court | High Court and Court of Session only |
| Are you in custody when you apply? | ✓ Yes, you are already arrested | ✗ No, you are still free |
| Triple test applies | ✓ Flight, tampering, witness influence | ✓ The same three questions |
| Common conditions | Surety, surrender passport, report to police | Cooperate with interrogation, no threats to witnesses, no leaving India without leave |
| Available for the gravest carve-out offences | Harder under PMLA, UAPA, NDPS twin conditions | Often barred or near-impossible under those statutes |
| Effect of media coverage on the test | ✗ Legally irrelevant | ✗ Legally irrelevant |
The single most consequential row is the last one, and it reads the same for both remedies. Whether you are asking for protection before arrest or release after it, the coverage does not enter the legal test. The court is meant to weigh the same three risks regardless of how loud the case has become.
If your problem is that an FIR has been registered against you and you are weighing pre-arrest protection against a challenge to the FIR itself, the parallel route of quashing is governed by its own standards, including the requirement that an FIR quashing order must state the allegations it is dealing with.
How Niyam helps you research bail law
Bail law looks simple in a slogan and turns dense the moment you open a real file. The principle is one line. The application runs through Gudikanti Narasimhulu on liberty, Sanjay Chandra on economic-offence gravity, P. Chidambaram on the triple test, Najeeb on delay defeating a statutory bar, Vijay Madanlal on the PMLA twin conditions, Antil on category-based directions, and Sisodia on prolonged custody, and which of those controls your matter depends entirely on your facts.
That is the work Niyam is built to speed up. Ask a question in plain English, such as “when does prolonged incarceration justify bail under UAPA” or “is gravity of offence enough to refuse anticipatory bail”, and Niyam answers with the relevant Indian judgments, every proposition cited to a real case you can open and read. No invented citations, no paraphrased law you cannot verify.
Two habits matter most in this area, and Niyam is designed around both. Before you cite a bail precedent, check whether it is still good law, because the bail decisions of 2022 have been read and re-read since. And when you need the holding of a long bail judgment fast, the approach in how to read a judgment helps you find the ratio without wading through the rest. If you are comparing tools for this kind of grounded research, the Niyam comparison page lays out how it differs from general chatbots that will happily fabricate a section number.
Indian legal research should be quick and anchored in real sources, not guesswork dressed up as confidence. In a field where a fabricated citation can sink a bail application in open court, that standard is the whole point.
Start for ₹100
Try Niyam on your next bail question. For ₹100 you get credits to run real research grounded in Indian judgments, with every answer cited to a case you can read. Create your account and start for ₹100.
Frequently asked questions
Does getting bail mean the court thinks the accused is innocent?
No. Bail decides only whether the accused waits for trial in custody or out of it. It is not a finding on guilt. The trial decides guilt, on evidence, later. A person on bail is still facing the full prosecution, and a person refused bail has still not been convicted of anything.
What is the triple test for bail?
It is the three questions a court asks in a non-bailable case: whether the accused is likely to flee, whether the accused might tamper with evidence, and whether the accused might influence witnesses. The Supreme Court applied it in P. Chidambaram v. Directorate of Enforcement (2019). It is a risk assessment about protecting the trial, not a preview of the verdict.
Does the seriousness of the crime decide bail?
Gravity is a relevant factor but it cannot decide bail on its own. In P. Chidambaram the Supreme Court held that the gravity of the offence alone cannot be a decisive ground to deny bail, and in Sanjay Chandra v. CBI (2012) it granted bail in a grave economic case, holding that pre-trial detention cannot be punitive. A serious charge raises scrutiny; it does not switch off the triple test.
Can a court refuse bail because the public is angry or the media is hostile?
Legally, no. Public sentiment and media coverage are not inputs into the bail test. In State of Maharashtra v. Rajendra Jawanmal Gandhi (1997) the Supreme Court called trial by press the “very antithesis of rule of law”. A bail decision should turn on flight risk, tampering and witness influence, regardless of how the case is being covered.
Why do famous accused sometimes stay in jail for a long time before bail?
Often because their cases fall under special statutes like the PMLA or UAPA, which attach “twin conditions” requiring the accused to show reasonable grounds that they are not guilty at the bail stage. That standard is hard to meet. But the Supreme Court held in Union of India v. K.A. Najeeb (2021) that these rigours melt away when incarceration is prolonged and the trial is not likely to finish in reasonable time, because Article 21 protects the right to a speedy trial.
What is the difference between bail and anticipatory bail?
Regular bail is sought after arrest, while you are in custody, and can be granted by a magistrate, Sessions court, High Court or Supreme Court. Anticipatory bail is sought before arrest, when you fear it, and only the High Court or Court of Session can grant it. Under the BNSS, regular bail in non-bailable cases falls under Sections 480 and 483, and anticipatory bail falls under Section 482.
What is the “bail is the rule, jail the exception” principle?
It is the settled position, traced to Justice V.R. Krishna Iyer in Gudikanti Narasimhulu (1978) and State of Rajasthan v. Balchand, that an accused person should ordinarily be released pending trial unless there are specific reasons to keep them in custody, such as flight risk or witness intimidation. The burden lies on the side arguing for jail, not on the accused arguing for liberty. It flows from the personal liberty guarantee of Article 21.
Can prolonged time in jail by itself get someone bail?
Yes, in appropriate cases. The Supreme Court has repeatedly held that long pre-trial custody with no realistic prospect of an early trial can justify bail on its own, because it turns detention into punishment without conviction. This reasoning drove the 2024 grant of bail in Manish Sisodia v. Directorate of Enforcement, where the accused had spent around 17 months in custody before the trial had even begun.
Are PMLA and UAPA bail conditions the same as ordinary bail?
No. They are stricter. The PMLA’s Section 45 and the UAPA’s Section 43-D(5) impose twin conditions that require the court to be satisfied there are reasonable grounds to believe the accused is not guilty and will not offend on bail. The Supreme Court upheld the PMLA conditions in Vijay Madanlal Choudhary (2022). These bars make bail harder, though constitutional courts can still grant relief where detention becomes excessive.
What happens if I am refused bail by the trial court?
A refusal is not the end. You can move the Sessions court or the High Court, and ultimately the Supreme Court, and a higher court can take a different view on the same record. Many high-profile bail grants come only after one or more refusals below. The appellate route is built into the system, which is why a single case can produce a sequence of bail orders.
Can media reporting of a case ever be restricted to protect a fair trial?
Yes, in limited circumstances. In Sahara India Real Estate Corporation v. SEBI (2012) the Supreme Court held that a person who genuinely fears that publicity will prejudice their fair trial can ask a constitutional court to postpone reporting of certain phases. Freedom of expression under Article 19(1)(a) is not absolute and can yield temporarily to the fair-trial right under Article 21.
Does the BNSS change how bail works compared to the old CrPC?
The core principles, the triple test and “bail is the rule”, carry over unchanged because they are constitutional, not merely statutory. What changed is the numbering and some procedure. Regular bail moved from Section 437 CrPC to Section 480 BNSS, the special powers of higher courts moved from Section 439 to Section 483, and anticipatory bail moved from Section 438 to Section 482.
How do I check whether a bail precedent is still valid?
Confirm it has not been overruled, doubted, or distinguished by a later bench before you rely on it. Bail jurisprudence moves quickly, and propositions from a few years ago are often read more flexibly today. Using a research tool that cites every proposition to a real, openable judgment, and checking the case’s current standing, protects you from building an argument on law that has shifted.