An employment bond is enforceable in India when it asks an employee to serve a minimum period or repay a genuine, documented cost the employer sank into them. It is void when it tries to stop the person from working at all. That distinction, drawn from Section 27 of the Indian Contract Act, 1872, is the whole game, and most bonds that land in court fail because the amount demanded looks like a punishment rather than a fair estimate of loss.
The Supreme Court settled the modern position on 14 May 2025 in Vijaya Bank & Anr. v. Prashant B. Narnaware, upholding a three-year service bond with a ₹2 lakh liquidated-damages clause. So the question for HR is no longer whether bonds work. It is how to draft one that survives a challenge, and how to enforce it without the recovery collapsing to a fraction of what the paper says.
The Contract Act Sections That Decide Everything
Every bond dispute in India runs through two provisions of the Indian Contract Act, 1872. Section 27 governs whether the restriction is allowed to exist. Section 74 decides how much money the employer can get back when someone leaves early.
The first rule, Section 27, says a person cannot be stopped from earning a living. A clause that blocks an ex-employee from joining a rival, or ties someone to a job they want to quit, is worthless the moment it is written, and no court will force anyone to keep working. What Section 27 allows is a promise to stay for a set number of years, or to repay the cost of training if the employee leaves before then. That is fair game, because it deals with time on the job, not freedom afterwards.
The second rule, Section 74, controls the amount. A bond can name a figure, say ₹2 lakh, but the employer only gets a reasonable sum, never more than the contract states. If that number is far bigger than what the company spent, a court treats it as a fine rather than repayment and cuts it down. That is why so many bonds pay out a fraction of the amount printed on them.
What The Supreme Court Actually Decided In 2025
The Narnaware ruling is the clearest recent statement of where bonds stand, and it went in the employer’s favour. Any HR team drafting or defending a bond should treat it as the controlling authority.
Prashant Narnaware joined Vijaya Bank as a probationary officer in 1999 and, in a 2006 recruitment drive, accepted a Senior Manager post carrying Clause 11(k): serve three years or pay ₹2 lakh on premature resignation. He resigned early to join IDBI Bank and challenged the clause. The Karnataka High Court struck it down; the Supreme Court reversed and upheld it in Vijaya Bank & Anr. v. Prashant B. Narnaware (2025 INSC 691).
The reasoning tracked six decades of precedent. A negative covenant that operates while the employment subsists does not offend Section 27; it perpetuates the contract for a set term rather than restraining future trade.
Leaning on the 1967 authority discussed below, the Court held that a minimum-service clause with proportionate liquidated damages is neither a restraint of trade under Section 27 nor void as against public policy under Section 23, and found no breach of Articles 14 and 19(1)(g). A ₹2 lakh figure against a senior banking role was treated as reasonable rather than penal.
The 1967 Foundation The Court Relied On
Narnaware did not invent the rule. It applied one the Supreme Court laid down in Niranjan Shankar Golikari v. Century Spinning and Manufacturing Co. (1967), the case every employment lawyer cites first.
Golikari was a shift supervisor trained for nine months in tyre-cord yarn manufacture under a five-year contract, who left for a competitor soon after. The Court upheld an injunction restraining him during the contract period and drew the line that still holds: a restriction operating during employment is generally not a restraint of trade, while one extending beyond it faces stricter scrutiny and is often struck as one-sided.
That split is why service bonds survive, and post-employment non-competes usually do not. TPB has covered the parallel question of whether non-compete clauses are valid in India, where the answer runs the other way.
Why Bonds Collapse: The Penalty Problem
A bond can be perfectly legal in structure and still recover almost nothing, and this is the trap most employers fall into. The failure point is Section 74, not Section 27. Courts consistently reduce the payable sum to the employer’s proven, proportionate loss.
The clearest illustration is M/s Sicpa India Ltd. v. Manas Pratim Deb (Delhi High Court, 2011). The bond demanded ₹2 lakh for early exit. Documented training cost came to roughly ₹67,595; the employee had already served two years, and the court awarded around ₹22,000 after apportioning the real expense against service rendered. The bond figure functioned only as a ceiling on what the court would award.
The contrast is Satyam Computer Services Ltd. v. Ladella Ravichandra before the Andhra Pradesh High Court, where the employer recovered because the training was documented and the bond reflected a quantified obligation rather than an arbitrary figure. The pattern across both cases is consistent: recovery follows evidence of actual spend, and evaporates without it.
| Case | Bond amount | Outcome | What made the difference |
| Vijaya Bank v. Narnaware (SC, 2025) | ₹2 lakh | Upheld in full | Reasonable sum, senior role, valid minimum-service clause |
| Satyam v. Ladella Ravichandra (AP HC) | Bonded sum | Recovered | Training documented and quantified |
| Sicpa India v. Manas Pratim Deb (Delhi HC, 2011) | ₹2 lakh | Cut to ~₹22,000 | Actual cost ~₹67,595, credit for service already given |
There is a wrinkle in how far this goes. The Section 74 text suggests a stated pre-estimate of loss need not be separately proved, yet cases such as VFS Global Services v. Suprit Roy have pushed employers to show real loss equal to the sum claimed. Courts hunt for training invoices either way, so unproven spend should not be counted on.
What Makes A Bond Actually Hold Up
Drafting decides enforceability long before a dispute reaches a judge. A bond built on the following elements is the kind courts have upheld; one missing them is the kind they reduce or ignore.
- Genuine, documented cost: Invoices for external training, certification fees, relocation, and specialist onboarding are what evidence a claim. The recoverable figure is the spend an employer can prove, so a bond untethered from real expense is a penalty waiting to be struck.
- Proportionate amount: The sum should track the actual investment rather than sit at a punitive multiple of it. A sliding scale that reduces as the employee serves out the term is far more defensible than a flat figure payable on day one or year three alike.
- Reasonable duration: Service periods of one to three years, calibrated to the size of the investment, are the ones courts have accepted. Open-ended or five-year-plus terms invite scrutiny.
- Free consent, no coercion: A bond signed under duress, or thrust at a new joiner minutes before they start with no chance to read it, is vulnerable. The Delhi High Court has held that coercion makes such instruments unenforceable.
- No restraint on future work: The clause has to stay confined to cost recovery. The moment it tries to stop the person joining a competitor or working elsewhere, Section 27 voids that part.
The new labour codes, operational from 21 November 2025, do not outlaw bonds, but their emphasis on transparent and fair terms of employment reinforces the same drafting discipline courts already demand.
The Retention Question Behind The Legal One
A bond is a legal instrument, but it answers a business problem: attrition and the cost of replacing trained people. Indian IT has lived this acutely, where firms such as TCS, Infosys, and Wipro run attrition in the double digits quarter after quarter, each departure carrying rehiring and re-training costs that bonds are meant to recoup.
The uncomfortable truth is that a bond is a weak retention tool. It raises the exit cost at the margin, but it does not make anyone want to stay, and an aggressively drafted one signals distrust to every new hire who reads it. Employers leaning hard on bonds while ignoring why people leave tend to find the bond recovers a modest sum and the talent goes anyway. Investment in real retention strategies does more to hold trained people than any liquidated-damages clause.
In The End…
Employment bonds are legal in India; the Supreme Court confirmed as much in May 2025, and they hold up when they recover a fair, documented cost rather than punish an exit. A few questions decide whether the bond an organisation is about to issue will survive that test.
Can the company produce receipts for the exact cost the bond recovers? Without them, the amount gets redrafted down to what can be proved, whether by HR now or a court later. Does the clause try to restrict where the person works afterwards? Any such restriction voids under Section 27 and can taint the rest of the instrument. Has the employee had a real chance to read and negotiate the terms, ideally in a document separate from the offer letter? Coercion is the fastest route to a clause no court will enforce.
A bond that clears those questions is an asset. One that fails them is a ₹2 lakh clause that recovers ₹22,000, at the cost of goodwill from everyone who signed it under protest.
FAQs
Are employment bonds legal in India?
Yes. A bond is enforceable when it asks an employee to serve a minimum period or repay a genuine, documented cost the employer incurred. It becomes void under Section 27 of the Indian Contract Act, 1872 only when it tries to stop the person from working elsewhere.
How much can an employer actually recover under an employment bond?
Only a reasonable sum backed by proof of spend, and never more than the figure named in the bond. Under Section 74, courts cut the amount down to the employer’s proven, proportionate loss, which is why many bonds pay out a fraction of the printed figure.
What did the Supreme Court decide in Vijaya Bank v. Narnaware (2025)?
On 14 May 2025 the Court upheld a three-year service bond carrying a ₹2 lakh liquidated-damages clause, holding that a minimum-service clause with proportionate damages is not a restraint of trade under Section 27.
Can an employment bond stop an employee from joining a competitor?
No. A clause restricting where someone works after they leave is a restraint of trade and is void under Section 27. A bond can recover cost, not block future employment.
What makes an employment bond hold up in court?
Genuine documented cost, an amount proportionate to that spend, a reasonable duration of one to three years, free consent with no coercion, and no restriction on future work.
