Citation analysis · 292,668 judgments

The second most-cited authority in Indian tax litigation is a land acquisition case

We read every ITAT judgment on this database and extracted every case it relies on — 732,794 citations in all. The results say something uncomfortable about what tax appeals are actually decided on.

BharatTax.net Research · 18 September 2026Figures computed from the BharatTax corpus. Text prepared with AI assistance.

Ask any practitioner to name the cases that matter in income-tax litigation and you will hear the same handful: Vegetable Products, Kelvinator, Malabar Industrial. Ask which case the Income Tax Appellate Tribunal actually reaches for most often, and nobody knows, because nobody has counted.

So we counted. Every one of the 292,668 judgments on this database was scanned for the authorities it cites, and the references resolved into decisions — merging parallel citations, so that 229 ITR 383 and 157 CTR 249 count as one case and not two. That yields 732,794 citation links across 26,763 distinct decisions.

The case at the top is no surprise. The case at number two should be.

Number two is not a tax case at all

Collector, Land Acquisition v. Mst. Katiji (1987) 167 ITR 471 (SC) is a two-page judgment about whether the State of Jammu & Kashmir should be forgiven four days’ delay in filing an appeal about compensation for acquired land. It contains no proposition of income-tax law whatsoever.

It is cited in 3,926 of our judgments.

Only National Thermal Power Co. Ltd. v. CIT (1998) 229 ITR 383 (SC) — on the Tribunal’s power to entertain a question of law raised for the first time — is cited more often, at 5,796.

The two together tell you what the Tribunal spends its time on. Neither is about how income is computed. Both are about whether the Tribunal will hear you at all: one about being late, the other about raising a point you did not raise below. Before the merits of a single appeal are reached, the threshold questions have already consumed more citation weight than any substantive provision of the Act.

That is not a criticism of the Tribunal. It is a description of where appeals are won and lost, and it is worth knowing before you draft your grounds.

The ten cases the Tribunal reaches for most

#CaseCitationForumCited in
1National Thermal Power Co. Ltd. v. CIT229 ITR 383SC 19985,796
2Collector, Land Acquisition v. Mst. Katiji167 ITR 471SC 19873,926
3CIT v. Kabul Chawla380 ITR 573HC 20163,235
4Malabar Industrial Co. Ltd. v. CIT243 ITR 83SC 20002,803
5CIT v. Vegetable Products Ltd.88 ITR 192SC 19732,642
6CIT v. Manjunatha Cotton & Ginning Factory359 ITR 565HC 20132,585
7CIT v. Multiplan India Ltd.38 ITD 320ITAT 19912,264
8CIT v. Reliance Utilities & Power Ltd.313 ITR 340HC 20092,114
9CIT v. Reliance Petroproducts Pvt. Ltd.322 ITR 158SC 20102,080
10Sumati Dayal v. CIT214 ITR 801SC 19951,964

Number of judgments in the corpus citing each decision. Parallel citations are merged, so a case reported in two series counts once.

Three of the ten are procedural rather than substantive: NTPC on new grounds, Katiji on delay, and Multiplan — a 1991 Tribunal decision on dismissing an appeal the appellant does not turn up to prosecute, cited 2,264 times. Of the rest, Kabul Chawla governs what an assessing officer may do in a search assessment where nothing incriminating was found, and Malabar Industrial sets the twin conditions for revision under section 263.

A thousand cases carry half the law

Reliance is far more concentrated than the size of the corpus suggests. Of the 26,763 decisions cited even once:

Top decisionsShare of all citations
Top 104.4%
Top 5012.5%
Top 10017.8%
Top 50036.0%
Top 1,00046.5%

Cumulative share of 666,656 attributed citations.

One thousand decisions — under 4% of those ever cited — account for nearly half of everything the Tribunal relies on. A practitioner who knew those thousand cases cold would recognise roughly one authority in two in any judgment put in front of them.

A quarter of the law in use predates 2000

Grouping every cited decision by the decade it was handed down:

1950s1.6%
1960s3%
1970s4.8%
1980s5.5%
1990s10.8%
2000s21.3%
2010s43.6%
2020s9.3%

The 2010s dominate, which is partly an artefact — our judgments run from 2015, and recent decisions cite recent decisions. The more durable figure is at the other end: roughly a quarter of all citations are to decisions handed down before 2000, and the 1970s alone are cited 30,485 times. Durga Prasad More (1971) 82 ITR 540 and Vegetable Products (1973) 88 ITR 192 are still doing daily work in assessments framed half a century after they were decided.

Reassessment is the engine

Counting judgments by the section in play, the machinery provisions outrank every charging provision:

SectionSubjectJudgments
143(3)Scrutiny assessment22,960
147Income escaping assessment16,924
250Procedure in appeal14,538
148Notice for reassessment13,255
144Best judgment assessment10,265
68Cash credits7,580
69AUnexplained money6,763
271(1)(c)Concealment penalty4,796

Judgments carrying a summary that identifies the section, 61,868 of the corpus.

Section 147 appears in 16,924 judgments and section 148 in 13,255 — largely the same appeals, since a reassessment carries both. The first genuinely substantive provision on the list is section 68, at 7,580. Whatever else the Tribunal is doing, it is mostly adjudicating how the Department got to the assessment rather than what the assessment contains.

And more than half of all judgments cite nothing at all

Of 292,668 judgments, 130,506 — 44.6% — cite a reported authority. The remaining 55% are decided without one: on the facts, on a concession, on a remand for verification, or by dismissal for non-prosecution.

This is the figure most at odds with how tax law is talked about. The reported cases we argue over govern a minority of what the Tribunal actually decides.

How this was done. Every judgment in the database was scanned for citation strings in the standard form — volume, reporter, page — across ITR, ITD, CTR, TTJ, SOT, Taxmann and the general series. Each reference was resolved to a decision, and parallel citations merged so a case reported in two series counts once.

What it does not do. “Cited” means a judgment mentions the authority. It does not distinguish a case followed from one distinguished, doubted or overruled — a heavily distinguished decision scores as highly as a faithfully applied one. The corpus is ITAT from 2015 onward together with the High Court and Supreme Court decisions those judgments cite; it is a picture of what the Tribunal relies on, not of Indian tax law entire. Citations are extracted mechanically, so OCR errors in the underlying orders survive into the counts.

Browse the full ranking on BharatTax Landmark Cases, where every decision has its own page with the judgments that rely on it.