Independent guarantee – To be precise: independent, yes — but a guarantee nonetheless

Note on Cass. com., 1 April 2026, no. 24-13.364, F-B

The independent guarantee is a singular figure among personal security interests. Belatedly codified at article 2321 of the Civil Code by Ordinance no. 2006-346 of 23 March 2006, it is the subject of a single provision. This may suggest that the legislator, having significantly framed the law of suretyship, intended in counterpoint to leave contractual freedom free to flourish in the field of independent guarantees.

The independent guarantee maintains an ambiguous relationship with the obligation it secures. Although it is autonomous — independent in principle — it cannot, however, fully emancipate itself from the underlying obligation, lest it cease to be a security at all. There is, indeed, no security without a debt to secure.

Hence a ridge to be walked: while the effectiveness of the independent guarantee must be preserved by enabling the creditor to be paid by the guarantor, autonomy must not be allowed to slide into arbitrariness.

This, in a few words, is what the Commercial Chamber of the Cour de cassation reminds us in a decision destined for the honours of the Bulletin, handed down on 1 April 2026 in the wake of the dispute arising from the aborted redevelopment of the Paris Gare du Nord (Cass. com., 1 April 2026, no. 24-13.364, F-B).

Before turning to the facts, let us set the stage by recalling the cardinal rules governing independent guarantees.

1. The applicable rules

Article 2321 of the Civil Code defines the independent guarantee as « the undertaking by which the guarantor obliges itself, in consideration of an obligation entered into by a third party, to pay a sum of money either on first demand or in accordance with agreed terms ».

The text, as can be seen, is both isolated and laconic.

It says little about the regime of the guarantee: « The guarantor is not bound in case of manifest abuse or fraud by the beneficiary or in case of collusion between the latter and the principal.

The guarantor cannot raise any defence based on the secured obligation.

Save where otherwise agreed, this security does not follow the secured obligation ».

It therefore fell to scholarship and case law to outline the regime, which we may attempt to sketch in broad strokes.

Autonomy. An essential element, the independent guarantee is… autonomous. Far from being a tautology, the assertion means that the guarantor enters into an undertaking distinct from that of the debtor. Fully distinct, since it commits to paying its own debt. To revisit a Roman-law distinction, what we find here is a duality of both obligatio (the power of constraint) and debitum (the debt). By contrast, in suretyship, although there are two ties of obligatio (the creditor has an action against both the surety and the debtor), there is only one debitum, since the surety pays the debtor’s debt. The independent guarantee is therefore autonomous in two respects.

« In consideration of an obligation entered into by a third party ». The independent guarantee is a security. As such, it strengthens the rights of a creditor. It follows that, if there is no claim, no obligation entered into by a third party, there can be no guarantee. The underlying debt is not the object of the guarantor’s obligation (he has his own debt); it is, however, its raison d’être. To allow the creditor to be paid by the guarantor when he has no claim against the debtor would betray the very notion of security and confer an unjustified enrichment upon the creditor.

« On first demand ». The guarantor must pay upon call, without being able to rely on defences drawn from the secured obligation. This is precisely what gives the independent guarantee its bite: pay first, discuss later — pay now, argue later. As in the dispute board mechanism, efficiency comes first.

The risk for the guarantor is therefore high: it may be wrongly called upon to pay, and the prospect of seeing the creditor sanctioned in court two or three years later offers little consolation. Case law therefore recognised, very early on, two safety valves: manifest abuse and manifest fraud (see in particular Cass. com., 11 December 1985, no. 83-14.457; Cass. com., 10 June 1986, no. 84-17.769, in the wake of the recognition of independent qualification by Cass. com., 20 December 1982, no. 81-12.579). To these two valves should be added collusion between the beneficiary and the principal, a hypothesis covered by article 2321 of the Civil Code but which, in practice, may also be regarded as a form of fraud.

Thus, in principle, it is very difficult for the guarantor to resist payment.

It nevertheless appears that another avenue of resistance exists, as evidenced by a recent decision of the Cour de cassation (Cass. com., 1 April 2026, no. 24-13.364, published in the Bulletin).

2. The facts

At the risk of simplifying them, the salient facts may be summarised as follows.

In February 2019, SNCF Gares & Connexions and Ceetrus Paganor (a subsidiary of New Immo Holding, hereinafter NIH) set up a single-purpose semi-public joint venture, the SEMOP « Gare du Nord 2024 », tasked with carrying out the redevelopment project of the Gare du Nord in Paris.

On 22 February 2019, SNCF G&C granted the SEMOP a concession to carry out the operation. Article 17.3 of the concession agreement provided that NIH would furnish a completion-bond independent guarantee in favour of SNCF G&C, in the amount of EUR 47 million, renewable annually (last renewal: 29 March 2021).

The « Gare du Nord 2024 » project encountered multiple difficulties — legal, financial and scheduling — and SNCF G&C ultimately notified, on 21 September 2021, the forfeiture of the concessionaire for serious misconduct, in application of article 52.1 of the concession agreement. The concessionaire SEMOP challenged the forfeiture before the administrative judge.

Without waiting for the outcome of that dispute, the creditor (SNCF G&C) called the guarantee provided by NIH, by registered letter dated 18 January 2022, for the full amount of EUR 47 million. The stakes were significant.

The guarantor, NIH, resisted the demand for payment and put forward three main arguments:

– the call did not relate to the object of the guarantee (the proper completion of the works of the abandoned « Gare du Nord 2024 » project);

– the call was manifestly abusive, since the beneficiary had renounced the very project for which the guarantee had been granted;

– the call was manifestly fraudulent, the alleged delays having been contractually extended by amendments.

The arguments were rejected by the Paris Court of Appeal, which ordered NIH to pay; NIH therefore appealed to the Cour de cassation.

3. The Cour de cassation’s solution

Let us say it from the outset: the solution does not change anything for NIH, whose appeal is dismissed and which will therefore have to pay the EUR 47 million. The practical interest for NIH is therefore limited.

On a theoretical level, however, the solution — published in the Bulletin — appears to be of some interest.

Indeed, the Cour de cassation lays down, with a clarity that we believe to be without precedent, a rule according to which « the independent guarantee cannot be called in respect of matters falling outside the object for which it was granted ».

The point is not to say that the guarantor may rely on a defence drawn from the principal relationship, but rather on a defence drawn from the very object of the independent guarantee contract. By way of comparison, an analogy may be drawn with insurance contracts: in order to determine whether the insurer must pay, one must verify the scope of cover as set out in the policy.

Second point: in order to verify whether the call of the guarantee is consistent with the object of the stipulated guarantee, the judge may, without disregarding the autonomy of the first-demand guarantee, refer « in addition to the guarantee itself, to the contract in consideration of which it was entered into ». Consultation of the underlying contract is therefore not prohibited; on the contrary, it is necessary in order to assess whether the call remains within the perimeter intended by the parties. The guarantee is autonomous, but it is not disconnected from the underlying contract.

The contribution of the decision appears significant in this regard. Alongside the means listed in article 2321 of the Civil Code (manifest abuse, fraud and collusion), another, judge-made one is added, drawn from the object of the guarantee. One could thus structure the means of resistance around two main axes: the call falling outside the object on the one hand, and abuse, fraud or collusion on the other — collusion being capable of being analysed as a particular instance of fraud.

Final point: the Cour de cassation maintains strict control over manifest abuse. The mere fact that the existence of the debtor’s debt has been challenged (here, before the administrative judge) does not suffice to render the call of the guarantee manifestly abusive or fraudulent. The rule is stringent for the guarantor: as long as the principal dispute has not been resolved, and as long as the beneficiary’s position remains defensible, the guarantor must pay.

It should be noted, in this regard, that the allocation of jurisdiction weighs upon the solution. The forfeiture of the SEMOP fell within the jurisdiction of the administrative judge, alone competent to assess the regularity of a decision taken under a concession agreement. As long as that judge has not ruled, the manifestly culpable nature of the creditor’s call cannot be regarded as established. NIH’s order to pay is, on this account, granted « subject to » any future order that the administrative judge may make against the SEMOP — a way of neutralising, as far as possible, the rigour of the autonomy principle.