Skip to content
ifscdesk.comSearch

How to check whether an IFSC code is still valid

If you are copying an IFSC code off an old cheque book, a saved payee, or a form you filled in years ago, there is a real chance the code no longer exists. Almost no lookup site will warn you — most will simply show it as though nothing has changed.

The quickest way to check is to paste the code into the search box on this site. If it has been withdrawn, the page says so at the top, tells you which release it was last present in, and — where we can establish it — names the code that replaced it.

The rest of this page explains how that check works, because you should not take a stranger's word about where your money is going.

Why a code stops existing

Two different things happen, and they are usually confused with each other.

Mergers. Several Indian banks have been amalgamated since 2015. Syndicate Bank went into Canara Bank, Vijaya and Dena into Bank of Baroda, Corporation and Andhra into Union Bank, Allahabad into Indian Bank, Oriental Bank of Commerce and United Bank into Punjab National Bank, Lakshmi Vilas into DBS. When that happens the old bank's codes are eventually withdrawn and its branches are renumbered under the acquiring bank.

Ordinary closures. Branches also close, merge into a neighbouring branch, or get renumbered without any bank-level merger at all. This is more common than people expect.

We hold every release of the RBI branch list that shipped a CSV since February 2016 — 135 of them, which reduce to 132 distinct builds once byte-identical rebuilds are removed. Comparing all of them against the current list gives 46,719 codes that were once in the list and are not in it now. Of those, roughly two thirds are merger-related. The other 15,712 are plain closures, including nearly 4,000 State Bank of India codes and about 1,700 Bank of India codes with no merger behind them at all.

That distinction matters. A page that tells you every withdrawn code "was merged into" something is inventing a reason for a third of them.

What "no longer valid" actually means here

We do not have a feed telling us when a code was switched off. What we have is every version of the official list, so what we can say precisely is: this code was in the release published on date A, and was not in the release published on date B. The change happened somewhere in that window.

That is narrower and more honest than a single invented cut-off date, and it is what the retired-code pages on this site show.

We are also deliberately slow to declare anything dead. A code is only marked retired once it has been missing from at least three consecutive releases and for at least 90 days. The reason is measurable: 9,324 codes in our archive disappeared from a release and later came back. A site that flagged a code the moment it vanished would be wrong thousands of times. At the time of writing, 38 codes are missing from the current list but have not yet cleared that waiting period, so we say nothing about them.

Three levels of answer, never mixed up

When a code is withdrawn, there are three quite different things we might be able to tell you, and running them together would be the easiest way to hurt someone.

1. We can name the replacement, with evidence

This applies to 15,319 codes. The evidence is one of three kinds, and every page records which one was used:

  • A mapping file published by the data provider itself. Old releases of the dataset carry old-to-new files that are no longer distributed on current releases. They hold 5,442 exact pairs covering Corporation Bank to Union Bank, Oriental Bank of Commerce to Punjab National Bank, and Lakshmi Vilas to DBS, and they resolve 5,067 of our retired codes — the rest name a successor that is itself no longer live. These are not inferences. Example: CORP0000001, the Corporation Bank branch at Udupi, maps to UBIN0900010.
  • An arithmetic rule published by the acquiring bank, checked against the branch name. Canara Bank's renumbering of ex-Syndicate branches follows a pattern, so SYNB0000100 (Hampankatta, Mangalore) becomes CNRB0010100 — and we only publish that when the branch name at the derived code also matches. More on this below, because it is where the danger is.
  • An exact address match inside the acquiring bank. 6,542 codes were resolved this way: the old branch's address string appears against exactly one live branch of the successor bank.

2. We know the bank merged, but not which code replaced yours

15,688 codes sit here. We know, for example, that VIJB0001001 was the Vijaya Bank branch at Adugodi, Bangalore, that it was last in the list published on 12 July 2021, and that Vijaya Bank was amalgamated into Bank of Baroda. We do not know which Bank of Baroda code now serves that branch.

So the page says exactly that, and lists Bank of Baroda's branches in the same city. It does not guess.

3. The code was simply withdrawn

15,712 codes. No merger, no successor on record. The page gives the archived branch record and the release it was last seen in, and says the branch was closed or renumbered. For a lot of people that is still the answer they came for: confirmation that the code is not merely mistyped.

Why one signal is never enough

This is worth spelling out, because it is the reason to trust the answers above.

The arithmetic rule for ex-Syndicate branches produces a live Canara Bank code for 93% of Syndicate's withdrawn codes — 4,118 of 4,443. It would have been easy to ship all of those: the derived code exists, it is a real Canara Bank branch, and nothing looks wrong.

But existence proves nothing. A rule that adds a constant will land on some live code almost every time, whether or not it is the right one, and there is no way to tell the two apart by looking at the result. So we require a second, independent signal before naming a successor.

In practice that is either the branch name at the destination matching the retired record, or an exact address match inside Canara Bank. Of the 4,118 derived codes, 3,710 are confirmed by the name and 323 more by an identical address where the name had changed. The remaining 406 have no second signal and are held back to level 2, where we say we do not know.

The address check earns its keep on exactly the cases the name check rejects. Canara renamed a lot of branches on renumbering: SYNB0008150, the Syndicate branch at Clock Tower, Ludhiana, became simply "Ludhiana" — but the address on both records is character-for-character identical, right down to Kachehari Road Near Clock Tower. The name gate cannot tell a renamed branch from a wrong one. Two signals can.

What to actually do with an old code

  1. Check it before you transfer, not after. Under the RBI's NEFT rules, credit to a beneficiary account is released solely on the account number, so the IFSC governs which branch the instruction is routed to, not who ultimately receives it. Getting it wrong is not a small clerical matter.
  2. If we name a replacement, still confirm it with your bank. Every retired-code page on this site shows the replacement code together with its branch name and address so you can sanity-check it yourself, and says to confirm before sending. That is not boilerplate. Data can be stale; the branch you want may have moved.
  3. If we cannot name a replacement, ask the successor bank. The acquiring bank knows which of its branches took over. Its own branch locator is the authoritative answer, and it is more current than any third-party file.
  4. Update the payee, do not just try the transfer. A withdrawn IFSC will usually be rejected outright, which is the good outcome. The bad outcome is a code that has been reissued to a different branch.

Why most sites will not tell you this

Not out of malice. A lookup site built from the current file simply has no way to know that a code ever existed — if it is not in today's list, the site has nothing to show, so it either returns nothing or, worse, keeps an old copy of the row and presents it as current.

We have seen both. Some sites still serve withdrawn codes with no warning at all, formatted identically to a live branch. If you are checking a code from an old cheque book, a page that looks completely normal is not evidence that the code works.

The only thing that makes a better answer possible is keeping the old versions of the file, which is a boring thing to do for years before it becomes useful.

Sources

  1. Razorpay IFSC dataset — every tagged release since February 2016 · https://github.com/razorpay/ifsc/releases (the archive this check is built on)
  2. Reserve Bank of India — NEFT frequently asked questions · https://www.rbi.org.in/Scripts/FAQView.aspx?Id=60
  3. ifscdesk.com data and method · https://ifscdesk.com/data-sources/ (the rules used to decide a code is retired)

Look something up

Figures on this page were measured against Razorpay IFSC release v2.0.62 (2026-09-01), which holds 183,214 branches, and against our archive of every published release since February 2016. See data and method for how we count.