There is a number that opens roughly every article ever written about Indian courts. It is in the crores, it is going up, and it is presented as self-evidently damning. It is also, on its own, almost information-free. A single aggregate covering three levels of the judiciary, every category of dispute, every stage of proceeding and every reason for delay cannot tell you anything actionable, and the reform proposals built on it tend to be as undifferentiated as the number itself.
This piece takes the number apart. Not to argue that the pendency problem is exaggerated, because it is not, but to show where it is a genuine crisis, where it is an ordinary queue, and where it is a definitional artefact. Those three things need different responses, and lumping them together is why so many reform proposals produce reports rather than results.
On the figures in this article
The National Judicial Data Grid publishes continuously updated pendency data, so any figure printed in an article is stale immediately. Magnitudes below are given to convey shape and proportion, not as citable statistics. For anything that matters, take the current figure from the Grid itself and record the date and view you took it from.
The First Cut: By Court Level
Split the total by level and the picture reorganises itself immediately. The overwhelming share of pending matters sits in the district judiciary, measured in crores. High Court pendency is measured in tens of lakhs. Supreme Court pendency is in the tens of thousands.
That distribution is the single most important thing to know about the crisis, and it is almost the exact inverse of where public attention goes. A reform that halved Supreme Court pendency would leave the national number essentially unchanged. A reform that improved district court throughput by a few percentage points would move it more than everything else combined.
It also reframes the judge strength argument. Vacancies in the High Courts are real, persistent and worth fixing, and they receive most of the coverage. Vacancies and capacity constraints in the district judiciary affect a docket two orders of magnitude larger, and receive a fraction of the attention.
The Second Cut: By Age
If you disaggregate only once more, do it by age. Pendency broken into age bands behaves very differently from pendency as a total, and the difference matters morally as well as analytically.
| Age band | What it usually represents | Is this a crisis? |
|---|---|---|
| Under one year | Matters moving through the ordinary steps: service, appearance, pleadings, framing of issues | No. This is a queue, not a backlog. Some of it is simply the time a proceeding takes. |
| One to three years | Matters at evidence, part-heard trials, matters awaiting a listing for final arguments | Partly. Length here reflects capacity and scheduling, and is where marginal improvements pay off. |
| Three to five years | Matters that have hit an obstacle: an unserved party, an absent witness, an interlocutory challenge | Yes, increasingly. By this point the delay has a specific cause that could usually be named. |
| Five to ten years | Matters stalled by stays, absconding accused, complex records or repeated transfers of the presiding officer | Yes. This is where the human cost concentrates and where targeted intervention is possible. |
| Over ten years | The old tail. Frequently stayed, frequently untraceable parties, frequently land and partition matters with large records | Yes, and it needs a different remedy from the rest. Generic productivity measures do not touch it. |
The distribution across those bands is heavily weighted toward the recent end. A large majority of pending matters are relatively young. The old tail is a minority of the total and it is where nearly all the genuine injustice sits.
Not all pendency is backlog, and the difference is the whole argument
A matter pending for eight months in a system with a heavy roll is a queue. A matter pending for twelve years is a failure. Counting them in the same total and calling the result a crisis produces reform aimed at the average case, which is not where the harm is. The right target is the tail, and the tail responds to different interventions from the body of the distribution.
The Third Cut: By Case Type
Break the district docket down by category and a small number of case types account for a disproportionate share. This is the disaggregation with the most direct policy implications, because a concentrated problem admits of a targeted solution.
- Cheque dishonour prosecutions. Proceedings under Section 138 of the Negotiable Instruments Act, 1881 are a category of their own by sheer volume. The Supreme Court took suo motu cognizance of the position in 2021, recording a pendency in the tens of lakhs and issuing directions aimed specifically at speeding these matters. They are largely documentary, largely repetitive, and structurally suited to procedural streamlining in a way most criminal matters are not.
- Motor accident claims. A very large civil block, with its own tribunal structure, its own quantum jurisprudence and a heavy dependence on insurers as parties.
- Land, title and partition disputes. Long by nature: many parties, old records, commissions, surveys, and frequently a parallel proceeding in a revenue forum over the same land.
- Execution petitions. A decree that has not been satisfied generates its own proceeding. Execution is separately registered, separately counted and notoriously slow, and it is largely invisible in reform discussion.
- Matters under special and local laws. Municipal, excise, labour and regulatory prosecutions in large numbers, often low in individual significance and high in aggregate volume.
- Matrimonial and maintenance proceedings. Emotionally heavy, procedurally protracted, and often accompanied by parallel proceedings in more than one forum.
Notice what this list implies. A substantial share of the district docket is not complex adjudication at all. It is high-volume, largely documentary work whose slowness reflects process design rather than judicial difficulty. That is good news, in the sense that process is easier to change than complexity.
Where the Number Is Definitional Rather Than Real
Some of the headline figure is an artefact of how cases are counted rather than a measure of unresolved disputes. This is not a defence of the system, it is a warning about the instrument.
Stayed matters
A matter stayed by a higher court sits in the pendency of the court below, which can do nothing about it. A meaningful share of very old district pendency is in this state, and attributing it to the trial court is simply wrong.
Execution counted separately
A suit decided and then executed generates two proceedings. This is correct as administration and misleading as a measure of disputes, because one dispute is counted twice.
Untraceable parties and absconding accused
Matters where a party cannot be served or an accused cannot be found remain pending indefinitely. They are not being delayed by anybody in the court.
Case unit inconsistency
What counts as one case varies with registration practice across States and case types. Aggregating them nationally produces a number whose unit is not stable, which is why cross-State comparisons need more care than they usually get.
Strip these out and the number gets smaller, but not small. The point is not that the crisis is invented. It is that a meaningful fraction of the total measures administrative bookkeeping rather than justice denied, and mixing the two makes it harder to see the part that is genuinely urgent.
The Flow View: Why Productivity Can Rise While Pendency Does
Pendency is a stock. Stocks are the accumulated difference between two flows: institutions and disposals. That elementary observation dissolves most of the apparent contradiction in reporting about Indian courts.
District courts dispose of an enormous number of matters every year, and the number has grown substantially over time. Pendency has also grown. Both are true, because filings have grown too. Litigation volume responds to economic activity, to enforcement policy, to legislative change and to how easy it has become to file. None of those are judicial variables.
The measure that captures this properly is the case clearance rate: disposals divided by institutions over the same period. Above one hundred per cent, the backlog shrinks. Below it, the backlog grows however hard everyone is working. It is a far more honest instrument than the total, and it is much less frequently quoted, presumably because it is much less alarming.
A court system can be more productive every single year and still fall further behind, if filings rise faster than disposals. That is not a failure of effort. It is arithmetic.
What Actually Drives the Old Tail
Since the tail is where the harm sits, it deserves its own diagnosis. The causes are known, are repeatedly documented, and are mostly not about judges.
Service of process fails
A very large volume of delay at the front end is simply the inability to bring a party before the court. Process serving is under-resourced, and every failed attempt costs a hearing date.
Witnesses do not appear
Repeated non-appearance, particularly of official witnesses such as investigating officers and medical witnesses, is among the largest single causes of criminal trial delay. It is an executive coordination problem presenting as a judicial one.
Interlocutory litigation interrupts the trial
Challenges to interim orders take the record upward and stop the trial clock. The more contested the matter, the more often this happens, and the cumulative effect on the timeline is severe.
Presiding officers are transferred
A trial part-heard by one officer and inherited by another loses continuity, and in some categories has to be substantially reheard. Transfer policy is an administrative choice with a direct throughput cost.
Adjournments are sought and granted
This is the one genuinely judicial lever, and it is real. The procedural codes limit adjournments, and the limits are honoured unevenly. It is a smaller factor than the rhetoric suggests and a larger one than the profession likes to admit.
Nobody is looking at the tail as a category
Old matters do not get old for one reason, and they are rarely managed as a distinct cohort with a distinct plan. Where courts have run targeted drives on the oldest matters, results follow, which suggests the problem is attention rather than possibility.
The Better Questions
If the argument of this article is right, then the useful contribution anybody can make to the pendency debate is to ask more specific questions. Here are the ones that would produce better reform than the headline number ever has.
What is the clearance rate, by court level and by State, and how has it moved? What share of pendency at each level is over five years old, and what are the recorded reasons? How much of the old tail is stayed by a higher court, and could those stays be reviewed systematically rather than case by case? What proportion of hearings are lost to failed service, and what would fixing process serving cost? How long does an execution petition take from decree to satisfaction? What happens to disposal rates in courts that ran a targeted old-cases drive, and did the effect persist?
Every one of those is answerable. Some are answerable from aggregate statistics on the National Judicial Data Grid. Others require case-level data, which is where the harder work is, because it means reading the record rather than counting it.
That distinction is the reason we build what we build. CourtMesh indexes the case record itself, across the Supreme Court, all twenty-five High Courts, the district judiciary and tribunals including NCLT, NCLAT, ITAT and CESTAT, drawn only from official government portals, with roughly 310 million records keyword-searchable and roughly 2 million carrying deeper semantic indexing. Aggregate dashboards tell you how many. Case-level data is the only thing that tells you what happened, and the pendency debate has been starved of the second for as long as it has been saturated with the first.
Move from counting cases to reading them
Aggregate pendency statistics answer system-level questions and nothing else. CourtMesh indexes the case record across the Supreme Court, all twenty-five High Courts, the district judiciary and tribunals including NCLT, NCLAT, ITAT and CESTAT, sourced only from official government portals. Search by issue, party, court, judge, act or section, look at what has actually been decided in a category of matter, and read the order at source.
Explore CourtMesh
