The short version

• The Liability Convention has two regimes. Absolute liability for damage on Earth or to aircraft in flight, fault liability for damage in orbit.
• It never defines fault, and there are no orbital rules of the road, which is why it has never been formally invoked for an on-orbit collision.
• Only one claim has ever been brought under it, by Canada against the USSR over Cosmos 954 in 1978, and it settled.
• In practice, debris risk is managed through licence conditions and insurance, not through the Convention.

In February 2009 a defunct Russian military communications satellite, Cosmos 2251, struck the operational Iridium 33 at a closing speed of around 11.7 kilometres per second. Both were destroyed. The event produced thousands of trackable fragments, many of which are still in orbit.

It was the clearest possible test case for the international liability regime governing collisions in space. No claim was brought.

Understanding why is the fastest route to understanding what the regime actually does.

Two regimes, one convention

The Convention on International Liability for Damage Caused by Space Objects entered into force on 1 September 1972 and had 100 states parties as at 1 January 2026. It gives content to Article VII of the Outer Space Treaty, and it splits damage in two.

Article II. Absolute liability. A launching state is absolutely liable to pay compensation for damage caused by its space object on the surface of the Earth or to aircraft in flight. No fault is required. If a stage lands on a farm, the launching state pays.

Article III. Fault liability. Where damage is caused elsewhere than on the surface of the Earth to a space object of another launching state, the first state is liable only if the damage is due to its fault or the fault of persons for whom it is responsible.

The logic is defensible. People on the ground did not choose to be exposed to falling rockets, so they get strict liability. Satellite operators voluntarily entered a shared environment, so they get a fault standard.

The word the Convention never defines

Article III turns on fault. The Convention does not say what fault means in orbit.

There is no duty to manoeuvre. There is no right of way. There is no rule about which of two converging objects should move, no obligation to maintain propulsion for collision avoidance, no standard for how often an operator must screen conjunctions, no defined probability threshold at which action is required, and no obligation to respond to another operator’s contact attempt.

Compare that with the sea. The COLREGs run to a detailed code of who gives way to whom in what circumstances. A court assessing fault in a shipping collision has a rulebook to apply. A Claims Commission assessing fault in an orbital collision would be inventing the standard as it went.

For Cosmos 2251 and Iridium 33 the difficulty was even starker. Cosmos 2251 was a dead object with no manoeuvring capability. It could not have moved. Fault, if any, would have to attach to launching a satellite without end-of-life disposal capability decades earlier, or to the quality of conjunction warning at the time. Neither is a comfortable claim, and neither state wanted to establish a precedent it would later be measured against.

Who counts as a launching state

The Convention defines a launching state four ways, and they can all apply at once. A state that launches, a state that procures the launching, a state from whose territory an object is launched, and a state from whose facility an object is launched.

Under Article V, where two or more states jointly launch, they are jointly and severally liable. A claimant can pursue whichever is most convenient, and the launching states sort out apportionment between themselves.

For a modern mission this can be a long list. A satellite manufactured in one country, owned by a company incorporated in a second, procured by an entity in a third, launched from the territory of a fourth on a vehicle operated by a company from a fifth. The Convention has no tie breaker.

The Registration Convention was supposed to make this tractable by requiring launching states to register objects with the UN. It has 77 parties and compliance for smallsats has been inconsistent, so the register is not a complete answer to “whose object is that”.

The one claim ever brought

In January 1978 Cosmos 954, a Soviet radar ocean reconnaissance satellite with a nuclear reactor on board, reentered over the Northwest Territories and scattered radioactive debris across a wide area of northern Canada.

Canada presented a claim to the USSR in 1979 for approximately CAD 6 million, covering the cost of locating and recovering the debris. It relied on the Liability Convention and on general principles of international law.

The USSR disputed the basis. The matter settled in April 1981 for CAD 3 million, in a protocol that did not admit liability under the Convention.

That is the entire case law. One claim, on the absolute liability side, settled without a determination.

What happens instead

The regime that actually manages debris risk is national licensing, and it has moved fast while the treaty layer has not.

• The FCC now requires post-mission disposal within five years for spacecraft in low Earth orbit below 2,000 km, replacing the old 25 year guideline. We set out the detail in our piece on that rule.
• The FAA imposes debris analysis requirements through Part 450 and financial responsibility through Part 440.
• ESA has moved to a five year expectation in its own debris mitigation policy, and the proposed EU Space Act would put debris mitigation and end-of-life disposal into a binding EU instrument.
• The IADC guidelines and ISO 24113 supply the technical baseline that most of these regimes draw on.

The enforcement point is worth emphasising. A debris mitigation plan filed with a licence application becomes an enforceable condition of that licence. That is how the FCC reached DISH Network in October 2023 over a geostationary disposal failure, in the first ever orbital debris enforcement action. We cover it in our case note.

Insurance does the rest. In-orbit cover responds to loss of the satellite, and third-party liability cover responds to claims by others, which for a US licensee is sized to a maximum probable loss determination.

Would a claim work today?

Suppose a functioning European satellite is destroyed tomorrow by a fragment traceable to a specific state’s object. Would a Liability Convention claim succeed?

The claimant state would need to establish, first, that the fragment came from the respondent’s object, which is genuinely hard for anything below about 10 centimetres. Second, that the respondent was at fault, against a standard that does not exist. Third, it would have to want the precedent, because the standard the Commission articulated would then apply to its own operators.

Article XIX(2) supplies the last disincentive. The Claims Commission’s award is final and binding only if the parties have agreed that it will be. Otherwise it is a recommendation.

None of this means the Convention is worthless. It supplies the strict liability rule that makes launch insurance pricing tractable, and it establishes that states are answerable, which shapes how they license. But anyone expecting it to function as a compensation mechanism for orbital collisions is reading a different document.