TL;DR: India has 5.39 crore cases pending across all courts as of 31 December 2025, according to figures the law ministry placed before the Lok Sabha in February 2026 from the National Judicial Data Grid (NJDG). The bulk sits in district and subordinate courts (4.76 crore), with 63.66 lakh in the 25 High Courts and 92,101 in the Supreme Court. The backlog has three main drivers: judicial vacancies running near 30 percent in High Courts and over 20 percent in the district judiciary, an adjournment culture that procedure has tried and failed to curb, and filings that keep outpacing disposals. Technology helps at the edges. eCourts Phase III, SUVAS translation, and the SUPACE research tool speed up paperwork, transcription, and language access, and AI legal research cuts the hours a lawyer spends finding the right precedent. None of it manufactures judges or hears arguments. This piece reads the NJDG honestly, separates the real causes from the convenient ones, and draws the line between what AI fixes and what it cannot.

On this page

What the NJDG actually counts

Before you trust any pendency number, you should know where it comes from. Almost every credible figure on India’s backlog traces back to one source: the National Judicial Data Grid, a public dashboard run under the eCommittee of the Supreme Court. The NJDG pulls live case data from the case information system installed in courts across the country, then aggregates pendency and disposal at the national, state, district, and individual court level.

That is what makes it useful. A litigant, a journalist, or a researcher can open the portal and see how many cases are pending in a particular court, how old they are, and how many were disposed in a given period, without filing a single RTI. The Department of Justice notes that case status for 23.81 crore cases and more than 23 crore orders and judgments is now searchable through the system. That is a genuine shift. Two decades ago, the size of the backlog was largely a matter of estimate and committee report.

The NJDG has limits worth naming. It counts cases, not disputes, so one matter that spawns multiple connected petitions can show up as several entries. It reflects what courts enter into the system, so data-entry gaps and classification differences across states do creep in. And a “pending” case is not the same as a stuck case, since some pendency is simply the normal time a matter takes to move from filing to hearing to judgment. Read with those caveats, though, the grid is the most honest mirror Indian justice has ever held up to itself.

The numbers: 5.39 crore and still climbing

Here is the headline, sourced to Parliament. On 13 February 2026, the Minister of State for Law and Justice, Arjun Ram Meghwal, told the Lok Sabha that based on NJDG data, over 4.76 crore cases were pending in district and subordinate courts as on 31 December 2025. Add the High Courts and the Supreme Court and the total comes to roughly 5.39 crore cases across the system.

The exact splits matter, so take them precisely. District and subordinate courts held 4,76,57,328 cases. The 25 High Courts together held 63,66,023. The Supreme Court held 92,101. None of these is shrinking. The Supreme Court’s pendency alone rose about 11.4 percent in two years, climbing from 82,674 cases at the end of 2023 to 92,101 at the end of 2025. High Court pendency rose 4.75 percent over three years, and the district judiciary’s load grew 5.84 percent in the same window.

The reason the pile keeps growing is arithmetic, not mystery. Filings outrun disposals. Tracking the Supreme Court month by month, the Supreme Court Observer found that in April 2025 the Court disposed of 3,485 cases while 4,029 were instituted, a clearance rate of about 86 percent. When a court clears fewer than 100 cases for every 100 that walk in, the backlog rises by definition, even when judges are working hard. That is the trap. The system can be busy and falling behind at the same time.

A second number puts the human cost in focus. As the government told Parliament, more than 71,000 cases have been pending in the High Courts for over 30 years. Across the district and High Courts together, the count of cases older than three decades runs past 1.8 lakh. Behind each of those entries is a person who filed expecting an answer in their working life and may not get one.

Where the cases pile up

The backlog is not spread evenly, and that changes how you read it. Around 88 percent of all pending cases sit in the district and subordinate courts. This is the base of the pyramid, the courts most people actually encounter, and the layer where delay does the most damage to ordinary litigants.

State-level concentration is just as stark. The Allahabad High Court carries the heaviest load of any High Court, with pendency above 12 lakh cases, followed by High Courts in Rajasthan, Bombay, Madras, and Madhya Pradesh, each holding several lakh. Allahabad’s burden is tied directly to its vacancy problem, which we will come to. A court running at half strength cannot clear a docket built for full strength.

The composition of the backlog matters too. Criminal cases dominate the district courts, and that feeds directly into India’s undertrial problem. The National Crime Records Bureau’s Prison Statistics India 2022 found that undertrials, people not yet convicted of anything, made up about 75.8 percent of the prison population, 4,34,302 of 5,73,220 inmates. Many sit in jail not because they were found guilty but because their trial has not finished and bail was out of reach. Slow trials and crowded prisons are the same crisis seen from two angles. If you want the framework courts are supposed to apply instead, it lives in bail is the rule, jail is the exception under Article 21.

There is also a quieter point the data makes. The single largest litigant in India is the government itself, across its departments and public-sector bodies. A large share of pending matters has the state on one side. That means the cause of much delay and a good part of the cure sit with the same institution that runs the courts.

Why cases pile up: vacancies, adjournments, procedure

Three causes do most of the work. The rest are mostly symptoms.

The first is vacancies. India simply does not have enough judges on the bench. As of mid-2026, the 25 High Courts had about 341 of roughly 1,121 sanctioned posts lying vacant, a shortfall near 30 percent, and the Supreme Court Observer found only three of the 25 High Courts functioning at full strength. The district judiciary runs a vacancy of around 20 percent. Step back to the population ratio and the picture is worse. India has roughly 15 to 21 judges per million people, far below the figure of 50 per million that the Law Commission recommended decades ago. PRS Legislative Research’s analysis of judicial vacancies traces how the collegium-and-government appointment process leaves seats empty for months and sometimes years. You cannot dispose of cases with chairs that have no one sitting in them.

The second cause is the adjournment culture. Hearings get pushed, then pushed again, and a matter that needs three sittings stretches across three years. Parliament tried to fix this by amending the Code of Civil Procedure to cap adjournments at three per party, and the Supreme Court upheld that cap in Salem Advocate Bar Association v. Union of India (2005). The Court also read in a sensible exception, that the limit cannot apply where the reason for delay is genuinely beyond a party’s control. In practice, the exception swallowed the rule. Adjournments remain routine in many courts, often granted as a matter of course, and the mandatory-cost provision meant to discourage them is rarely enforced with teeth.

The third cause is procedure itself. Indian litigation moves through a long sequence of interlocutory applications, service of notice, framing of issues, evidence, and appeals, with each stage open to challenge. The new criminal codes try to attack this with statutory timelines, for example fixed periods for filing charge sheets and pronouncing judgment, an approach laid out in the new criminal laws under BNS, BNSS and BSA. Whether those timelines hold or quietly slip is one of the open questions of the next few years.

None of these three is glamorous. There is no single villain. The backlog is what you get when too few judges work through too much procedure while postponements pile on, year after year.

What pendency costs the country

Delay is not only a private misfortune for the litigant. It is a drag on the whole economy, and the government’s own analysis has said so plainly.

The Economic Survey 2018-19 devoted a chapter to the lower judiciary and reached a striking conclusion. It estimated that a 100 percent case clearance rate, meaning courts dispose of as many cases as come in and stop the backlog from growing, could be achieved in the lower courts by adding only about 2,279 judges, a number already within sanctioned strength. The fix, in other words, was largely about filling existing posts rather than inventing new ones. The Survey framed contract enforcement and case backlog as among the biggest constraints on doing business in India, and put the value of the reform bluntly: “given the potential economic and social multipliers of a well-functioning legal system, this may well be the best investment India can make.”

That framing matters because it reorders priorities. A pending commercial dispute freezes capital. A pending land matter blocks investment. A pending criminal trial keeps a presumed-innocent person in jail and a family without income. Various estimates have put the cumulative economic cost of judicial delay at more than 2 percent of GDP. You do not need to trust a single number to see the logic. Justice that arrives a decade late is, for most economic purposes, justice that did not arrive.

This is also why the pendency debate keeps circling back to capacity. Better technology, faster procedure, and more alternative dispute resolution all help. But the Economic Survey’s point stands: the cheapest, most direct lever is filling the judge’s chair.

What technology can really do

Now to the part everyone wants to talk about, and where the hype runs hottest. Technology has a real role in attacking pendency. It is just a narrower role than the headlines suggest.

Start with the honest version of the claim. Technology does not decide cases, and it does not create judges. What it does is remove friction from the work around a case, the searching, summarising, translating, scheduling, and transcribing that eats a lawyer’s and a registry’s day. Cut that friction and you free human time for the part only humans can do.

Case triage is one example. Software can flag which matters are ready for hearing, which are stuck waiting on a single missing document, and which are duplicates or already-settled disputes clogging the list. Scheduling tools can build cause lists that group similar matters and reduce dead time. Live transcription, already running in the Supreme Court’s Constitution Bench hearings, produces a record without a stenographer falling behind. Translation, which we cover below, opens judgments to litigants who do not read English.

Then there is legal research, which is where AI bites hardest for the working lawyer. Finding the controlling precedent, checking whether it is still good law, and reading how recent benches applied it used to take hours per question. Tools built on Indian case law now compress that to minutes. The former Chief Justice of India, D.Y. Chandrachud, called AI in legal research a “game-changer” offering “unmatched efficiency and accuracy” at the Indo-Singapore Judicial Conference in April 2024. That is the realistic promise. A lawyer who researches in a tenth of the time files faster, drafts tighter, and wastes less of the court’s hours, and across thousands of lawyers that adds up. How this differs from a generic chatbot is the subject of native legal AI for India versus generic GPT.

There is a second front that does not involve AI at all: keeping cases out of court in the first place. Mediation, Lok Adalats, and arbitration divert disputes that never needed a judge. The Mediation Act, 2023 pushes pre-litigation mediation for civil and commercial matters, and Lok Adalats settle lakhs of cases in single sittings. Diversion is unglamorous, but every dispute resolved outside the courtroom is one that never joins the NJDG count.

eCourts, SUVAS and SUPACE: the state’s own AI push

The judiciary is not waiting for the private sector. It is building its own digital spine, and the scale is real.

The backbone is the eCourts project. In September 2023 the Union Cabinet approved eCourts Phase III with an outlay of ₹7,210 crore over four years. The aim is paperless, digitised courts, universal e-filing and e-payments, digitisation of legacy records, and e-Sewa Kendras so that litigants without computers can still file and check status. This is the plumbing on which everything else runs. Without digitised records, there is nothing for any AI tool to read.

On top of that sit two named AI systems. The first is SUVAS, the Supreme Court Vidhik Anuvaad Software, which translates judgments from English into regional languages. The numbers are not small. As the Supreme Court has reported through government channels, tens of thousands of judgments have been machine-translated, with over 42,000 rendered into 17 regional languages and tens of thousands more into Hindi, now accessible through the e-SCR portal. For a litigant in a Tamil or Bengali district court, a Supreme Court ruling that was effectively locked behind English becomes readable. That is access to justice in the most literal sense. The free judgment portal it feeds is explained in e-SCR and neutral citations.

The second is SUPACE, the Supreme Court Portal for Assistance in Court’s Efficiency, launched in April 2021. SUPACE reads case files, extracts relevant facts and documents, and helps judges and their researchers manage the paperwork of a case. What it deliberately does not do is decide. At the launch, then Chief Justice S.A. Bobde was explicit that the system would not be used for decision-making and that judges’ autonomy and discretion would be retained. The design philosophy is augmentation, not automation. Years on, SUPACE remains largely in pilot use rather than full national rollout, a reminder that building courtroom AI is slower than announcing it. The wider story of these tools is covered in how AI is being used in Indian courts, and the rules now governing such tools in the Supreme Court’s AI rules for India.

What AI does not fix

Here is the part the marketing skips. AI attacks the symptoms of pendency, not its root. Be clear-eyed about the limits, because pretending otherwise is how good tools get blamed for failing at jobs they were never meant to do.

AI does not appoint judges. The single biggest cause of the backlog is empty benches, and no model fills a sanctioned post. AI does not grant adjournments or refuse them, so the adjournment culture survives whatever software the registry installs. AI does not hear oral arguments, weigh credibility of a witness, or exercise the discretion that the law reserves for a human judge. Chandrachud, the same Chief Justice who praised AI research, was equally firm that technology can augment human capability but not replace human judgment.

There is a sharper risk specific to legal AI: confident fabrication. A general-purpose chatbot will, when it does not know, invent a case name and a citation that look entirely real. Indian courts have already encountered fake citations generated this way, a problem we examine in AI-hallucinated citations in India. In a field where a wrong citation can sink a brief or mislead a court, a tool that guesses is worse than no tool at all. This is why the duty to verify sits squarely on the lawyer, not the model, and why checking whether a precedent is still good law is not optional.

Finally, technology can widen gaps as well as close them. A litigant with a digitised, well-resourced court and a lawyer using modern research tools moves faster. A litigant in a court still entering data by hand, with counsel who has none of these tools, does not. The promise of eCourts is genuine, but its benefit is uneven until the plumbing reaches every district. Technology is a multiplier. It multiplies whatever capacity already exists, which is exactly why capacity, judges and infrastructure, has to come first.

AI and pendency, side by side

It helps to see the division of labour in one frame. The backlog has structural causes and procedural ones. AI touches some and is powerless over others.

Lever against pendencyHelped by AI / tech?What actually moves the needle
Finding the right precedent fast✓AI legal research grounded in Indian case law
Translating judgments into regional languages✓SUVAS and machine translation on e-SCR
Transcribing and recording proceedings✓Live transcription, already in use
Triage, scheduling, duplicate detection✓Case-management software and cause-list tools
Drafting and summarising long files✓Assistive tools like SUPACE for registries and judges
Filling vacant judge posts✗Faster appointments, the Economic Survey’s core fix
Curbing the adjournment culture✗Judicial discipline and enforced cost rules
Hearing arguments and deciding cases✗Human judges, by constitutional design
Diverting disputes out of court✗ (mostly)Mediation, Lok Adalats, arbitration

Read the table the right way. The ✓ rows are where a lawyer or a court saves hours today. The ✗ rows are where the real backlog lives, and they need institutional reform, not a better app. Anyone selling AI as the cure for pendency is selling the left column as if it were the right.

How Niyam helps you work despite the backlog

You cannot fix India’s vacancy rate from your desk. What you can fix is how much of your own time the backlog wastes, and that is where Niyam is built to help.

The largest hidden tax on a litigating lawyer is research time. When every pending matter competes for your attention, the hours spent hunting for the controlling judgment, reading around it, and checking whether it still holds are hours you do not have. Niyam answers a plain-English question, such as “what is the test for bail in economic offences” or “is mandamus available against a government-aided college,” with the relevant Indian judgments, every proposition cited to a real case you can open and read. That is the difference between AI legal research built for India and a generic chatbot that guesses.

Three habits make the backlog hurt less. First, before you rely on any precedent, check that it is still good law, because an overruled judgment cited in a crowded court is a self-inflicted wound. Second, when a judgment runs to a hundred pages, use judgment summarisation to find the ratio fast, then read the key passages yourself. Third, to stay current as the Supreme Court draws new lines month to month, follow the June 2026 Supreme Court digest.

Niyam will not clear the NJDG. No tool will. What it does is make sure that the part of justice you control, the quality and speed of your own legal research, is never the reason a matter drags. To see how it compares with the way you research today, look at the comparison.

Start for ₹100

Put your next research question to Niyam. For ₹100 you get credits to run real research grounded in Indian judgments, with every answer cited to a case you can read and verify. Create your account and start for ₹100.

Frequently asked questions

How many cases are pending in India right now?

About 5.39 crore cases are pending across all courts as of 31 December 2025, based on NJDG data the law ministry placed before the Lok Sabha in February 2026. That breaks down into roughly 4.76 crore in district and subordinate courts, 63.66 lakh in the 25 High Courts, and 92,101 in the Supreme Court.

What is the NJDG?

The National Judicial Data Grid is a public dashboard run under the eCommittee of the Supreme Court that aggregates pendency and disposal data from courts across India. It lets anyone see how many cases are pending at the national, state, district, and individual court level, and how old they are, without filing an RTI.

Why are there so many pending cases in India?

Three main reasons. Judicial vacancies run near 30 percent in the High Courts and over 20 percent in the district judiciary. An adjournment culture stretches hearings out despite a legal cap of three adjournments. And new filings consistently outpace disposals, so the backlog grows by arithmetic even when judges work hard.

Which court has the most pending cases?

Among the High Courts, the Allahabad High Court carries the heaviest load, with pendency above 12 lakh cases. In absolute terms, the district and subordinate courts together hold the overwhelming majority of all pending matters, around 88 percent of the national total.

How many judges does India have per million people?

India has roughly 15 to 21 judges per million population, well below the figure of about 50 per million that the Law Commission recommended decades ago. Filling sanctioned but vacant posts is the single most direct way to raise that ratio.

Can AI reduce pendency?

AI helps at the edges, not at the root. It speeds up legal research, translation, transcription, case triage, and drafting, which frees human time. It cannot appoint judges, grant or refuse adjournments, hear arguments, or decide cases. The structural causes of the backlog need institutional reform, not software.

What is eCourts Phase III?

It is the third phase of the national eCourts project, approved by the Union Cabinet in September 2023 with an outlay of ₹7,210 crore over four years. It aims to build digital, paperless courts, universal e-filing and e-payments, digitised legacy records, and e-Sewa Kendras for litigants without computer access.

What is SUVAS?

SUVAS, the Supreme Court Vidhik Anuvaad Software, is an AI tool that translates judgments from English into regional languages. Over 42,000 judgments have been translated into 17 regional languages, plus tens of thousands into Hindi, now accessible through the e-SCR portal.

What is SUPACE?

SUPACE, the Supreme Court Portal for Assistance in Court’s Efficiency, is an AI tool launched in April 2021 that reads case files, extracts relevant facts and documents, and helps judges and researchers manage paperwork. It does not decide cases. At launch, the Chief Justice stressed that judges’ autonomy and discretion would be retained.

Does AI decide cases in Indian courts?

No. AI tools in Indian courts are assistive. They help with research, translation, transcription, and file management, but the decision is reserved for a human judge by constitutional design. The Supreme Court’s own AI tools were built explicitly to augment, not replace, judicial decision-making.

How long are the oldest cases pending?

Some have been pending for more than 30 years. The government told Parliament that over 71,000 such cases sit in the High Courts alone, and across the district and High Courts together the count exceeds 1.8 lakh cases older than three decades.

What is a case clearance rate?

It is the ratio of cases disposed to cases filed in a given period. A clearance rate of 100 percent means courts dispose of as many cases as come in, so the backlog stops growing. India’s clearance rate has often run below 100 percent, which is why pendency keeps rising.

How does pendency affect undertrial prisoners?

Heavily. Because criminal trials drag, a large share of prison inmates are undertrials who have not been convicted. The NCRB’s Prison Statistics India 2022 found undertrials made up about 75.8 percent of the prison population, many held simply because their trial had not finished.

What does judicial delay cost the economy?

A great deal. The Economic Survey 2018-19 called case backlog and weak contract enforcement among the biggest constraints on doing business in India, and estimates have put the cumulative cost of delay at more than 2 percent of GDP. The Survey argued that fixing the lower judiciary may be the best investment India can make.

Will technology alone solve the backlog?

No. Technology multiplies whatever capacity already exists, so it helps fast courts get faster and resourced lawyers work harder. But it cannot create judges, enforce discipline on adjournments, or hear cases. The cheapest and most direct fix the government’s own analysis identified is filling vacant judicial posts.