Skip to content
Perspectives

ReferenceInternational Business Law

Why your limitation of liability in B2B contracts fails

Almost every B2B contract limits liability, and under German law almost none of those clauses holds. The evidence, the case law behind it, and the four levels where limiting liability actually works.

Peter Poleacov20 min

Aerial view straight down onto a dam: the still reservoir on one side, the controlled release of water on the other.

In 2014, the German Federal Ministry of Justice had more than 1,200 companies surveyed about their contracting practice. Two results stand out. First: for most companies, limiting their own liability is important or very important; among large companies, more than 80 percent say so. Second: only a minority considers the legal room for doing so insufficient.1 The two results only make sense together once you know the third finding. Lars Leuschner, the study’s author, later put it soberly: on the basis of the Federal Court of Justice’s case law, effective limitations of liability are “nearly impossible”.2 German contracting practice relies, across the board, on clauses the law does not give it. It simply does not notice. The worst case is rare, and most B2B disputes end in confidential arbitration.3

After twenty years of drafting and negotiating international contracts, I consider this gap between expectation and legal reality the most underrated contract risk in German law. This article shows why the standard liability clause fails, what the case law leaves standing, and on which four levels limiting liability actually works.

The starting point: unlimited liability

Absent a contractual arrangement, a debtor is liable in damages for any breach of duty it is responsible for (Section 280 BGB). Responsibility already attaches to simple negligence (Section 276 BGB). Liability is unlimited in amount and in principle covers consequential losses: production downtime, business interruption, lost profits. Picture a supplier whose component, worth a few hundred euros, shuts down a production line. Under the statutory default, that supplier is liable for the entire standstill loss. Statutory liability ceilings are rare exceptions (for instance Section 10(1) of the German Product Liability Act, Section 431 HGB). This is exactly why every seller, contractor, and service provider wants to limit that liability by contract. And this is exactly where the problem begins.

Almost everything is standard terms

The first error concerns the scope of judicial review. Standard terms are all terms pre-formulated for repeated use that one party imposes on the other (Section 305(1) BGB). The courts read that broadly: templates, text modules, and even industry-standard clauses suffice. The escape route of “individual negotiation” is far narrower than almost everyone believes. Negotiating individually means more than negotiating. The user must genuinely put the core of the liability clause itself up for discussion — and offer unlimited statutory liability as a real alternative.4 Which company does that in plant engineering, IT outsourcing, or the supply industry? Even intensely negotiated large-volume contracts consist almost entirely of standard terms by this yardstick. Courts assess each clause on its own: fighting over the price does not turn the untouched liability clause into an individually negotiated term. Details in “No individual agreement”.

These errors are measurable. In the ministry’s study, 32 percent of respondents believed the parties could simply agree that the liability clause counts as individually negotiated. Nearly half of the respondents from large companies believed intense contract negotiations were enough.1 Both are wrong — the first has even been decided expressly by the Federal Court of Justice.5

What content review leaves standing

For the review itself, the specific clause prohibitions of Sections 308 and 309 BGB do not apply directly between businesses. But the Federal Court of Justice treats them as indicative.6 The result is a mandatory core in three parts. No standard term may touch it:

  • liability for intent and gross negligence, including that of vicarious agents (Section 309 No. 7 lit. b BGB);
  • liability for injury to life, body, and health (Section 309 No. 7 lit. a BGB);
  • liability for the breach of essential contractual obligations (cardinal duties). For simple negligence, this liability may only be capped in amount — and not below the foreseeable, contract-typical damage.7

This covers not only outright exclusions but any indirect relief. A general shortening of the limitation period fails under Section 309 No. 7 BGB, because it indirectly eases liability for gross fault (BGH, judgment of 22 September 2015 – II ZR 340/14). The same fate meets no-fault liability extensions in purchasing terms (BGH, judgment of 5 October 2005 – VIII ZR 16/05 “Baumarkt”). And the popular rescue formula “as far as legally permissible” does not help: an overreaching clause is not trimmed down to the permissible maximum, it is struck.

So what remains? The Federal Court of Justice has sketched the architecture of a clause that holds:

This clause holds. But it holds because it concedes what the statute demands anyway. The liability risks that matter in practice — quality deviations and delay — almost always concern essential contractual obligations. And the consequential losses one actually wants to exclude — above all the customer’s lost profits — are precisely what is contract-typical and foreseeable in commercial dealings. The customer buys the performance in order to earn money with it.2 The foreseeability formula hands back with one hand what it withholds with the other. The Anglo-American standard exclusion of “consequential damages” has its own version of this trap — see “Consequential damages”.

The four levels where limitation works

The conclusion is not resignation but a different division of labor. Limiting liability happens on four levels, and the clause is only one of them — the weakest. Whoever wants to exhaust the options ends up at the advanced move.

First level: the subject matter, meaning the statement of work

What is not owed cannot be breached. The statement of work is not subject to the strict review of content (Section 307(3) BGB). A precise, exhaustive description of the owed performance — defined metrics, tolerances, acceptance criteria — limits liability more effectively than any liability clause. If you do not promise 99.9 percent availability, you are not liable for 99.9 percent. It is the most powerful drafting tool German law offers, and it is chronically neglected: the statement of work is the most important contract topic nobody writes about.

In standard terms the room is narrow, but it exists. Use the BGH architecture above, plus a monetary cap for the segment of simple negligent breaches of essential duties. The cap must cover the contract-typical damage. Aligning it with the coverage of the business liability insurance has proven itself. That amount can be justified as reasonable, and it is actually funded when things go wrong. Whoever wants more must genuinely negotiate. Put the core of the clause up for discussion, give the customer a real choice (say, unlimited liability at a higher price), and document the process. That is laborious, uncomfortable, and the only path to a clause that may do more than the statute. And whoever sits in a supply chain aligns the tiers: your own upstream supplier must be liable to you to the same extent you are liable to your customer. Otherwise the difference stays on your books.

Third level: the governing law

The benchmark of content review is not a constant of nature. It follows from the applicable law, and that is a matter of drafting. Under the CISG, liability clauses are, on the better view, measured against the Convention’s flexible framework rather than the strict BGB model. A genuine limitation, at least for simple negligence, becomes achievable — one of the strongest reasons not to exclude the CISG reflexively. Swiss law reviews standard terms only vis-à-vis consumers (Article 8 of the Swiss Unfair Competition Act). Between businesses what remains is essentially a surprising-clause check. The only mandatory limits are those of Article 100 of the Swiss Code of Obligations: no exclusion for intent and gross negligence. A limitation of liability for simple negligence that German law prohibits in standard terms is routine under Swiss law. The precondition is an international setting; in purely domestic cases mandatory German law stays applicable (Article 3(3) Rome I Regulation). The third lever, the choice of forum, is important enough to get a level of its own.

For the other side of the table, the inversion holds. It is a tactical instrument in its own right: as a buyer, you benefit from Germany’s strictness. Choosing German law makes the limitation clauses in your supplier’s terms of sale collapse wholesale. You can then quietly accept your supplier’s limitation clause and enjoy your latte.

Fourth level: the forum, meaning the arbitration clause and the seat

Whoever drafts the liability clause decides only half the case. The other half is decided by whoever reads it later. An arbitral tribunal is a private court. It is bound by the law, but not by the case law of the Federal Court of Justice, and nobody reviews whether it applied the law correctly: there is no révision au fond, an error of law leads to annulment only exceptionally, and the sole gateway is public policy (Section 1059(2) No. 2 lit. b of the German Code of Civil Procedure, Zivilprozessordnung, ZPO).

This invites a misconception I meet regularly: an arbitration clause does not make the liability clause valid. If German law governs, Sections 305 et seq. BGB apply in arbitration too. According to the only solid study on the point, a survey of experienced arbitration practitioners, the standard-terms objection is raised in roughly a third of arbitrations governed by German law, in well over 90 percent of those cases against limitations of liability — and the majority of tribunals follow the cardinal-duties case law of the Federal Court of Justice.8

The advantage lies elsewhere, and it can be stated precisely: arbitral tribunals are more generous about what counts as individually negotiated. What the Federal Court of Justice refuses to treat as an individual agreement, tribunals frequently accept where the parties negotiated intensely and the clause was textually changed from the first draft.8 The arbitration clause shifts the threshold of content review; it does not switch the review off. Negotiate hard, document it, and an arbitral tribunal may recognize the individual agreement you would lose before a German court.

On top of that comes what often weighs more in practice than the doctrine: the proceedings are not public, the parties select the arbitrators themselves, and the award is easier to enforce across the 172 contracting states of the New York Convention than a German judgment. The price: a single instance, no appeal, higher costs.

One technical point many overlook: the arbitration clause is an agreement independent of the main contract (Sections 1029(2), 1040(1) sentence 2 ZPO). If the main contract is invalid or is avoided, the arbitration clause regularly survives.9 The clause that decides your liability outlives the contract it sits in. All the more reason to draft it deliberately: institution, number of arbitrators, seat, language of the proceedings, applicable law. Each of these is a switch, not a formality.

The advanced move: a foreign seat plus opting out of standard-terms law

The idea: choose German law, but carve the Sections 305 to 310 BGB out of it, and combine that with an arbitration clause seated outside Germany, as a rule in Switzerland. I say up front that the construction is not settled.

The first objection arrives immediately and sounds fatal: Sections 305 et seq. BGB are mandatory law, not at the parties’ disposal, as the Federal Court of Justice has expressly held.5 But the objection only covers one half. It says you cannot contract out of standard-terms law within German law, that is, by private autonomy. It says nothing about which law governs in the first place. That question is answered by party autonomy, meaning conflict of laws. Party autonomy can do what private autonomy cannot.9

Before state courts this leads nowhere: under Article 3(1) Rome I Regulation you can only choose a national legal system, not a tailored extract from one. In arbitration, a different conflict-of-laws regime applies. Under Section 1051(1) sentence 1 ZPO the tribunal decides according to the rules of law designated by the parties — and rules of law are not the same thing as a legal system. Swiss law at the seat permits the same in Article 187(1) of its Private International Law Act. German law without Sections 305 to 310 BGB is therefore a choosable benchmark, at least where the transaction is genuinely international.10

The Swiss seat is not ornamental here; it carries two loads. First, it covers the residual risk that the Rome I Regulation reaches arbitral tribunals after all, which the prevailing view denies — but the Court of Justice of the European Union has never ruled on it.11 Second, and this is the real breaking point of the whole construction: for the arbitration agreement itself no tailored benchmark can be chosen, only a national law. Choose German law for it, and you expose the arbitration clause of all things to review under Section 307 BGB. Choose Swiss law, which knows no open B2B content review, and that lever disappears.10

That leaves the heaviest objection: is standard-terms law part of public policy? If it were, the opt-out would run empty, because public policy is the limit of every choice of law. The most prominent advocate of the opt-out is, of all people, a former judge of the Federal Court of Justice, and he denies it: a selective waiver of Sections 305 to 310 BGB sets aside nothing that the founding principles of the legal order treat as indispensable. The minimum protection against unreasonable disadvantage survives anyway, because a correction of pre-formulated clauses oriented on the demands of justice still takes place after the opt-out, to the extent public policy requires it.9 He therefore considers the opt-out effective even in purely domestic cases.12

And the Federal Court of Justice? It had the question before it in January 2025: German law, a DIS arbitration clause, seat in Berlin, and the parties’ express declaration to waive the application of Sections 305 to 310 BGB. It upheld the arbitration agreement, because that agreement is independent of the fate of the main contract, and expressly left open whether the waiver itself is effective. Its dictum is nevertheless the most important sentence on the topic: regardless of whether the standard-terms provisions belong to public policy, their non-application may in the individual case produce a result that offends it — namely where the tribunal upholds a contractual arrangement whose formation can no longer be understood as an expression of contractual self-determination, or which leads to plainly intolerable consequences.13

That sets the frame. The opt-out shifts the review; it does not abolish it. What remains is public policy, Section 242 BGB and, where German law governs the arbitration agreement, review of the arbitration clause itself. What disappears is the cardinal-duties casuistry, not the reasonableness test.

My line in advising clients: this construction belongs in contracts with a genuine international element, with a counterparty of equal standing, and with a liability regime that would still pass as reasonable without the crutch of the opt-out. For purely domestic contracts it is a leap into the dark: whether Article 3(3) Rome I Regulation bites there is undecided, and the dissenting voices carry weight.12 Anyone who writes the opt-out into standard terms of sale for every customer has misunderstood the instrument. In the end they risk both: the liability clause, and the arbitration clause meant to carry it.

What you should do

  1. Audit your terms. Take your terms of sale and service. If the liability clause contains a blanket exclusion of consequential damages, or a cap without the mandatory carve-backs, it is very likely invalid — and the unlimited statutory liability applies.
  2. Rebuild the clause on the BGH architecture. Fault-based tiers, the cardinal-duties carve-back, a cap only in the permissible segment, aligned with insurance coverage. No fantasy exclusions that endanger the whole clause.
  3. Treat the statement of work as a liability instrument. Precise, exhaustive, measurable. This is where the drafting freedom lies that content review cannot touch.
  4. Use the governing-law level in international contracts. Keep and shape the CISG deliberately; where the negotiating position allows, consider Swiss law with an arbitration clause.
  5. Decide the forum, don’t inherit it. The arbitration clause is not a closing formality but a liability decision: institution, seat, number of arbitrators, language, and applicable law all need to be set deliberately. In high-value contracts with an international element, consider opting out of standard-terms law — but only with the right law governing the arbitration agreement, and only where the liability regime stays reasonable without it.
  6. Reconcile insurance and residual risk. What cannot be limited legally must be insured or priced in. A liability clause is no substitute for risk management.

Frequently asked questions

Can liability be limited in standard terms under German law?

Only within narrow limits. Liability for intent and gross negligence, for injury to life, body, and health, and for the breach of essential contractual obligations (cardinal duties) cannot be excluded; for simple negligence, liability for foreseeable, contract-typical damages must remain intact. What can be freely limited is essentially only liability for simple negligent breaches of non-essential duties. The commercially relevant risks are precisely the ones that standard terms cannot cap.

Does Section 309 No. 7 BGB apply between businesses?

In substance, yes. The specific prohibitions of Sections 308 and 309 BGB do not apply directly in B2B contracts, but the Federal Court of Justice treats them as indicative. Exclusions of liability for gross fault or for injury to life, body, and health therefore regularly fail in B2B standard terms as well, via Section 307 BGB.

Is intense negotiation enough to make the liability clause an individually negotiated term?

No. Negotiating a term individually means more than negotiating: the user of the clause must genuinely put its core content up for discussion, offering the statutory regime of unlimited liability as a real alternative. Negotiations about other parts of the contract do not radiate onto the liability clause, and a signed confirmation that the clause was individually negotiated is legally ineffective.

Is a monetary liability cap permitted in German standard terms?

Only in a narrow segment. A cap may neither touch liability for gross fault nor undercut liability for foreseeable, contract-typical damages from the breach of essential obligations. It remains permissible for simple negligent breaches of essential duties if it covers the contract-typical damage, for example aligned with the coverage of the user’s business liability insurance.

What role does the CISG play for limitations of liability?

A major one. Where the CISG applies, liability clauses are, on the view I share and advocate, measured against the Convention’s flexible framework rather than the strict BGB model. A genuine limitation of liability, at least for simple negligence, becomes achievable. For exporters this is one of the strongest reasons not to exclude the CISG.

Does an arbitration clause defeat German standard-terms review?

Not in the way it is usually claimed. If German law governs, an arbitral tribunal applies Sections 305 et seq. BGB as well, and the majority of tribunals follow the cardinal-duties case law of the Federal Court of Justice. The solid advantage is a different one: tribunals are more willing than state courts to treat a clause as individually negotiated where the parties negotiated intensely and the clause changed from the first draft. The arbitration clause shifts the threshold of review; it does not switch the review off.

Can German standard-terms law be excluded?

Not within German law, because Sections 305 et seq. BGB are mandatory. Through conflict of laws it can: in arbitration, Section 1051(1) sentence 1 ZPO allows the parties to choose rules of law rather than an entire legal system, and therefore German law without Sections 305 to 310 BGB. The construction is combined with a seat outside Germany, usually in Switzerland, because only a national law can govern the arbitration agreement itself. The Federal Court of Justice expressly left the question of effectiveness open in 2025, and review via public policy remains in place either way.

Notes

  1. Leuschner/Meyer, final report of the research project “AGB-Recht für Verträge zwischen Unternehmen”, commissioned by the Federal Ministry of Justice, 2014. Results summarized in Leuschner, NJW 2016, 1222 (1224 f.): limiting liability rated important by 81 and 89 percent of large companies; 32 percent believe the individually-negotiated status can simply be agreed; 40.6 and 47.4 percent consider intense contract negotiations sufficient. ↩ ↩2

  2. Leuschner, Die Kontrollstrenge des AGB-Rechts, NJW 2016, 1222 (1224): effective limitations of liability are “nearly impossible” (“nahezu unmöglich”) on this basis. ↩ ↩2

  3. On the more liberal tendencies of arbitral tribunals Leuschner/Meyer, final report (fn. 1), pp. 137 et seq., 146, 148; Leuschner, NJW 2016, 1222 (1225). ↩

  4. Settled case law, e.g. BGH, NJW 2013, 856 (“Bring or Pay”): no individual negotiation despite extensive negotiations and legal review, because the core content was not genuinely put up for discussion. Summarizing Leuschner, NJW 2016, 1222 (1223). ↩

  5. BGHZ 200, 326 = NJW 2014, 1725: the application of Sections 305 et seq. BGB is not at the parties’ disposal; a mutually signed “individually negotiated” confirmation is ineffective. ↩ ↩2

  6. BGH, judgment of 19 September 2007 – VIII ZR 141/06, BGHZ 174, 1 = NJW 2007, 3774; settled case law on the indicative effect of Sections 308, 309 BGB between businesses. ↩

  7. Foreseeability formula: settled case law, e.g. BGH, NJW 1985, 3016; NJW 1993, 335; summarizing Leuschner, NJW 2016, 1222 (1223 f.). ↩

  8. Like state courts, arbitral tribunals are bound by the law, but not by the case law of the Federal Court of Justice: Wittuhn/Quecke, NZG 2014, 131. The empirical basis is a survey of the members of the German Arbitration Institute (190 usable responses), not an analysis of awards, which are not published: Leuschner/Meyer, SchiedsVZ 2016, 156 (158 et seq.). The standard-terms objection was raised in 31 percent of the proceedings governed by German law, in 96 percent of those cases against limitations and exclusions of liability; 62 percent of respondents said tribunals follow the cardinal-duties case law. On the more generous understanding of individual negotiation ibid. (160): intense negotiations (76 percent) and a textual change from the first draft (69 percent) frequently suffice for arbitral tribunals; see also ICC, interim and partial award of 29 January 2001 – arbitration no. 10279, SchiedsVZ 2005, 108. ↩ ↩2

  9. Sections 305 et seq. BGB are mandatory as a matter of substantive law, but they are not overriding mandatory provisions within the meaning of Article 9 Rome I Regulation; their exclusion by way of conflict of laws therefore remains possible: Leupertz, RInPrax 2025, 2; Ostendorf, ZIP 2022, 730 (731). On public policy Leupertz, ibid.: the parties’ selective waiver of Sections 305 to 310 BGB does not collide with public policy. The author is a former judge of the Federal Court of Justice. ↩ ↩2 ↩3

  10. Section 1051(1) sentence 1 ZPO speaks of “rules of law”, not of a legal system; Article 187(1) of the Swiss Private International Law Act permits the same at a Swiss seat. On the whole question including the law governing the arbitration agreement Ostendorf, ZIP 2022, 730 (732 et seq.), noting that for the arbitration agreement itself German law minus Sections 305 et seq. BGB cannot be chosen; further Leuschner/Rieländer, AGB-Recht im unternehmerischen Rechtsverkehr, 2021, Rechtswahl paras. 57 f.; Pfeiffer, NJW 2012, 1169 (1170) and NJW 2025, 866; Kästle, NZG 2014, 288 (295); Zons, RInPrax 2025, 172. Against the effectiveness of an opt-out von Westphalen, NJW 2025, 2216. The law governing the arbitration agreement follows, by analogous application of Article V(1)(a) of the New York Convention, primarily the chosen law and, failing that, the law of the seat (BGH, order of 9 January 2025 – I ZB 48/24; BayObLG, order of 5 June 2024 – 102 SchH 42/24e). ↩ ↩2

  11. The Rome I Regulation does not apply to arbitration agreements (Article 1(2)(e) Rome I Regulation); that it does not govern the substance in arbitration either follows, on the prevailing view, from the fact that private arbitral tribunals are not “courts of a Member State” (Nueber, SchiedsVZ 2014, 186 [188 et seq.], citing ECJ, Nordsee and Denuit). There is no ruling of the Court of Justice on the Rome I Regulation in arbitration, and the opposing view is gaining ground (see Ostendorf, ZIP 2022, 730 [732]). ↩

  12. In favor of effectiveness even in purely domestic cases Leupertz, RInPrax 2025, 2, and Pfeiffer, NJW 2012, 1169; likewise Zons, RInPrax 2025, 172. More cautious Ostendorf, ZIP 2022, 730 (735), who requires a genuinely international element; against Schulzweida/Deger, IHR 2025, 185, and Stoffels, AGB-Recht, 4th ed., para. 1076. There is no case law on the point. ↩ ↩2

  13. BGH, order of 9 January 2025 – I ZB 48/24, para. 42 (the effectiveness of the exclusion is expressly left open): regardless of whether the standard-terms provisions form part of German public policy, the recognition or enforcement of an award may — because the standard-terms provisions were not applied in the arbitration — lead to a result that offends public policy, for instance where the tribunal upholds a contractual arrangement whose formation can no longer be understood as an expression of contractual self-determination, or where a contractual arrangement leads to plainly intolerable consequences. ↩

Reference: Poleacov, P. (2026). Why your limitation of liability in B2B contracts fails. INN.LAW. https://inn.law/en/perspectives/limitation-of-liability/