Dr. Rakesh Kumar Gupta v. State of Uttar Pradesh (2026): can a referee judge on a split verdict reopen unanimous findings of guilt?

When two judges of a High Court Division Bench hearing a criminal appeal disagree, Section 392 CrPC provides the mechanism for resolving the deadlock: the appeal is laid before a third judge, and it is then disposed of according to the opinion of the judges who heard it including that third judge. A recurring question has long lurked beneath that procedure. If the two original judges agreed on most of the case and split only on one accused or one point, is the referee judge confined to the narrow point of disagreement — or may that judge reopen the whole appeal, including the findings the first two judges were unanimous about? In Dr. Rakesh Kumar Gupta v. State of Uttar Pradesh, decided on 9 June 2026, a two-judge Bench of the Supreme Court doubted the answer the Court had given a quarter-century earlier and sent the question to a larger Bench. What follows is a digest of a reference — a question framed, not a question answered.

The facts in brief

The matter traces back to a 1991 incident registered at Police Station Wazirganj, Lucknow, in Uttar Pradesh. Three members of the Rastogi family — described in the reports as Anil, Ajay and Atul Rastogi — were tried and convicted by the Sessions Court for offences under Sections 302 read with 149 and Section 148 of the Indian Penal Code, 1860 (murder in furtherance of an unlawful assembly's common object, and rioting armed with a deadly weapon) and sentenced to imprisonment for life.

On appeal, a Division Bench of the Allahabad High Court split. The two judges were unanimous in affirming the conviction of two of the accused, but they differed on the culpability of one of the co-accused. Because of that single point of difference, the appeal — heard as one composite appeal for all the accused — went before a third judge under Section 392 CrPC.

The referee judge did not stop at the point of dissent. According to the reports, that judge proceeded to re-examine the entire case and disturbed the findings against the two accused whose convictions the first two judges had unanimously upheld. It was this exercise of power — a third judge reopening findings on which the original Bench had agreed — that brought the matter to the Supreme Court in connected criminal appeals filed by the complainant, Dr. Rakesh Kumar Gupta, and by the State of Uttar Pradesh.

The question referred

The central problem the Bench identified is whether the words of Section 392 CrPC — which speak of "the appeal" being decided according to the opinion of the judges including the third — confine the referee judge to the point on which the original judges disagreed, or open up the whole appeal for fresh consideration. On the reports, the Bench framed a cluster of questions around that core, including:

• Whether a referee judge under Section 392 CrPC must choose between the two conflicting opinions of the original judges, or may instead craft an independent third opinion;
• Whether a referee judge may disturb unanimous, concurrent findings of guilt when the disagreement between the original judges existed only as to one co-accused's culpability; and
• Whether, where the referee judge disagrees with concurrent findings, the proper course is to refer the matter onward to a larger Bench rather than to decide it.

The reports indicate that the Bench answered the first of these for itself but reserved the second and third — the questions about disturbing unanimous findings — as the core of the reference for the larger Bench. The Bench also flagged a constitutional dimension: whether allowing a referee judge to reopen unanimous findings simply because the co-accused happened to file a single composite (joint) appeal produces an arbitrary disparity, under Article 14, as against accused persons who file separate appeals and whose unanimous convictions would not be exposed to the same reopening.

What the Court held

It is important to be precise about what the two-judge Bench did and did not do. It did not lay down a binding rule on the scope of a referee judge's powers. What it did was express doubt about the existing authority and refer the question for authoritative determination.

The authority doubted is Sajjan Singh v. State of Madhya Pradesh, (1999) 1 SCC 315, understood as supporting the proposition that a third judge under Section 392 CrPC may examine the entire case afresh. The Bench in Rakesh Kumar Gupta expressed serious reservations about the correctness of that view. Its tentative inclination — stated as the reason for doubting Sajjan Singh, not as a final ruling — was that a referee judge's role ought to be confined to the points of dissent rather than treated as a licence to review the entire case.

The Bench's concern was both textual and structural. Textually, it questioned whether the statutory language of Section 392 of the 1973 Code (and the way it differs from the corresponding Section 429 of the older Criminal Procedure Code, 1898) really authorises a wholesale re-hearing. Structurally, it worried that letting a third judge reopen unanimous findings makes the fate of a settled conviction turn on the accident of how the appeals happened to be grouped — a disparity it doubted could be reconciled with equality before the law.

Having recorded those doubts, the Bench referred the question of whether Sajjan Singh states the correct law, and the associated questions on the scope of a referee judge's jurisdiction, to a larger Bench. That is the operative outcome: a reference, leaving the substantive question open.

Analysis

The significance of Rakesh Kumar Gupta lies in the tension it exposes between two competing intuitions about appellate procedure.

The first intuition favours finality and parity. Where two judges have agreed — reached a concurrent, unanimous finding of guilt — that agreement looks like the strongest kind of appellate conclusion, settled by two minds rather than contested between them. On this view, a third judge brought in only because of a disagreement about a different accused has no warrant to disturb what was never in disagreement. To allow otherwise is to let an accused's exposure to acquittal-or-reopening depend on the procedural happenstance of whether they were tried and appealed jointly or separately — which is precisely the Article 14 concern the Bench raised.

The second intuition treats the appeal as an indivisible whole. Section 392 speaks of "the appeal" being disposed of according to the opinion of the judges including the third; on a literal reading, the third judge is hearing the appeal, not a severed fragment of it, and is therefore entitled — perhaps obliged — to form a complete view. This is the reading associated with Sajjan Singh, and it has the virtue of avoiding artificial line-drawing between which findings a referee judge may and may not revisit within a single composite proceeding.

The reference does not resolve that tension; it sharpens it. Until the larger Bench speaks, practitioners face genuine uncertainty about how far a third judge may go. The cautious course, after Rakesh Kumar Gupta, is to assume the question is live: a litigant relying on Sajjan Singh for the proposition that a referee judge may re-open the whole case can no longer treat that proposition as settled, and a litigant resisting such re-opening now has a Supreme Court reference recording doubt to point to.

It is also worth noting what is not in play here. Because the underlying prosecution dates to a 1991 incident, the applicable code is the Criminal Procedure Code, 1973, and the relevant provision is its Section 392. The reports do not indicate that the Bench engaged the corresponding provision of the Bharatiya Nagarik Suraksha Sanhita, 2023, which has since replaced the 1973 Code for new matters; the doctrinal question, however, is structural and would carry over to any successor provision framed in similar terms.

Why it matters

For criminal appellate practice, Rakesh Kumar Gupta is a marker that a settled-seeming rule is now open. The scope of a referee judge's jurisdiction under Section 392 CrPC affects the security of every conviction that survives a split verdict in the High Courts: if the larger Bench confirms that a third judge may reopen unanimous findings, then a single point of dissent can place an entire composite appeal back in issue; if it confines the referee judge to the point of disagreement, then unanimous concurrent findings acquire a finality that the accident of joint appeals cannot unsettle.

The case is a reminder, too, of how a reference functions in the Indian system. A two-judge Bench that doubts an earlier decision of co-ordinate or higher strength does not overrule it; it records its doubt and refers the question upward, leaving the impugned authority standing until a larger Bench decides. The practical takeaway for now is one of caution: Sajjan Singh has been doubted but not displaced, and the law on the powers of a referee judge after a split verdict awaits authoritative settlement.

Related on Valkya

• Mukesh Kumar Yadav: appellate sentencing
• Munna Moyuddin Shaikh: modification of a life sentence
• State of Tripura v. Panna Ahmed: the limits of Section 311 CrPC recall

Sources

• Verdictum, "Scope Of Reference To Third Judge U/S 392 CrPC Limited To Points Of Dissent? Supreme Court Refers Sajjan Singh Judgment To Larger Bench" — https://www.verdictum.in/supreme-court/dr-rakesh-kumar-gupta-v-state-of-uttar-pradesh-2026-insc-632-refers-sajjan-singh-judgment-to-larger-bench-1615725
• Verdictum, "Weekly Overview | Supreme Court Judgments: June 09 – June 12, 2026" — https://www.verdictum.in/weekly-summary/supreme-court-judgments-june-09-june-12-2026-1615930
• LiveLaw, "Dr. Rakesh Kumar Gupta v. State of Uttar Pradesh & Ors — 2026 LiveLaw (SC) 615" — https://www.livelaw.in/amp/sc-judgments/2026-livelaw-sc-dr-rakesh-kumar-gupta-v-state-of-uttar-pradesh-ors-537446
• Supreme Court Observer, "SCO.LR | 2026 | Volume 6 | Issue 3" — https://www.scobserver.in/journal/sco-lr-2026-volume-6-issue-3/