The short version

• Nuclear power in space is lawful. The Outer Space Treaty bans nuclear weapons in orbit, not nuclear power sources, and missions such as NASA’s Perseverance rover already run on plutonium-238.
• The main international text is the UN’s 1992 Principles Relevant to the Use of Nuclear Power Sources in Outer Space (resolution 47/68). It is not binding, and parts of it, including a rule that reactors use only highly enriched uranium, no longer match how the United States is designing reactors.
• In the US, launches are approved under NSPM-20 of 20 August 2019, which sorts missions into three tiers by radiological risk. Commercial launches go through the FAA, which published Advisory Circular 450.45-1 on space nuclear systems in October 2023.
• Policy is moving fast. An executive order of 18 December 2025 calls for a lunar surface reactor ready for launch by 2030, and on 8 October 2026 NASA and the Department of Energy signed an agreement to work together on space nuclear power and propulsion.

On 8 October 2026 the NASA Administrator and the US Energy Secretary signed a memorandum of understanding on space nuclear power and propulsion. NASA’s release says the agreement takes effect on 1 November 2026, names a nuclear electric propulsion mission, SR-1 Freedom, planned for 2028, and a future lunar fission surface power system called LR-1. It quotes Jared Isaacman as saying the agency is “entering the ‘Nuclear NASA-era'”.

That raises a question I am asked often: is it even legal to put a nuclear reactor in space? It is. But the rules are older, softer and less coherent than most people assume. This explainer sets out the treaty position, the 1992 UN Principles, the US launch approval system and the places where they no longer fit together.

Two kinds of space nuclear power

The law treats two technologies differently, so it helps to separate them. Radioisotope power systems produce heat and electricity from the natural decay of a radioactive isotope, usually plutonium-238. They are not reactors and cannot go critical. The US Department of Energy describes how Perseverance’s Multi-Mission Radioisotope Thermoelectric Generator converts heat from the decay of plutonium-238 into electrical power, with an operational life of 14 years.

Fission reactors split uranium in a controlled chain reaction. They can produce far more power, which is why they are the focus of lunar base and propulsion plans. They also carry different risks: a reactor that has operated holds fission products that are much more radioactive than its fresh fuel.

What the Outer Space Treaty says, and does not say

The 1967 Outer Space Treaty does not mention nuclear power. Article IV prohibits placing “in orbit around the Earth any objects carrying nuclear weapons or any other kinds of weapons of mass destruction”, a rule I discussed in our piece on weapons in orbit. A reactor or a radioisotope generator is not a weapon, so Article IV does not apply.

The articles that matter are the general ones. Article VI makes States internationally responsible for national activities in space, including those of companies. Article VII makes launching States liable for damage. Article IX requires due regard for other States and the avoidance of harmful contamination. Our Outer Space Treaty explainer goes through each of them.

The 1992 UN Principles

The specific rules are in the Principles Relevant to the Use of Nuclear Power Sources in Outer Space, adopted by the UN General Assembly without a vote on 14 December 1992 as resolution 47/68. As a General Assembly resolution, they are recommendations, not a treaty. Their preamble limits them to power sources “devoted to the generation of electric power on board space objects for non-propulsive purposes”.

Principle 3 is the core. It says nuclear power sources should be restricted to missions “which cannot be operated by non-nuclear energy sources in a reasonable way”. It sets a radiation protection goal for accidents of 1 mSv a year for individuals. It allows reactors only on interplanetary missions, in a “sufficiently high orbit” where fission products can decay before re-entry, or in low Earth orbit if the reactor is moved to a sufficiently high orbit after use. Reactors “shall use only highly enriched uranium 235 as fuel” and “shall not be made critical before they have reached their operating orbit or interplanetary trajectory.” Radioisotope generators must have containment designed to survive re-entry and impact without scattering radioactive material.

Principles 4 to 9 deal with procedure and consequences. A launching State must carry out a thorough safety assessment before launch and make the results public. If a nuclear-powered object malfunctions with a risk of re-entry, the State must notify the States concerned and the UN Secretary-General and keep updating them. After re-entry, the launching State must offer help with location and clean-up. Principle 9 confirms that liability follows the Outer Space Treaty and the Liability Convention, and that compensation includes the costs of search, recovery and clean-up.

Principle 11 said the Principles would be reopened for revision within two years. The text UNOOSA publishes today is still the 1992 version.

Why these rules exist: Cosmos 954

The Principles were negotiated in the years after the Soviet satellite Cosmos 954, which carried a nuclear reactor, re-entered over northern Canada on 24 January 1978 and scattered radioactive debris. Canada’s statement of claim relied on the Liability Convention and Article VII of the Outer Space Treaty and sought C$6,041,174.70. Under a protocol of 2 April 1981, the Soviet Union paid C$3,000,000 in full and final settlement. Principle 9’s express inclusion of clean-up costs reflects that case. I covered the claim in more detail in our Liability Convention explainer.

The 2009 Safety Framework

The second international instrument is the Safety Framework for Nuclear Power Source Applications in Outer Space, jointly prepared by the Scientific and Technical Subcommittee of the UN Committee on the Peaceful Uses of Outer Space and the International Atomic Energy Agency, and presented in 2009 as document A/AC.105/934. It is a management framework rather than a set of limits: it asks governments to have a justification and authorisation process, and asks developers to manage safety through design, testing and mission planning. It says plainly that it “provides voluntary guidance and is not legally binding under international law.” A working group of the Subcommittee has continued to review how the Framework and the 1992 Principles are applied.

How the United States approves a nuclear launch

US practice is set by National Security Presidential Memorandum 20 of 20 August 2019. It sets safety guidelines expressed as probabilities: an accident exposing a member of the public to between 25 millirem and 5 rem should have a probability of no more than 1 in 100, between 5 and 25 rem no more than 1 in 10,000, and above 25 rem no more than 1 in 100,000.

It then sorts launches into three tiers. Tier I covers radioactive sources up to 100,000 times the IAEA’s A2 transport value, and the sponsoring agency head approves the launch. Tier II covers larger quantities, higher-risk Tier I cases and any reactor using low-enriched uranium (below 20 percent uranium-235), and requires review by an Interagency Nuclear Safety Review Board. Tier III covers missions with a 1 in 1,000,000 or greater chance of an accident exposing someone to more than 25 rem, and any reactor using fuel other than low-enriched uranium. Government Tier III launches need presidential authorisation, which the Director of the Office of Science and Technology Policy may give for non-defence missions.

For commercial missions, NSPM-20 makes the Secretary of Transportation the licensing authority in all three tiers. The FAA responded with Advisory Circular 450.45-1, “Launch and Reentry of Space Nuclear Systems”, in October 2023. It explains how applicants under the Part 450 regime, which I described in our Part 450 explainer, should present radiological safety analysis.

Where the rules no longer fit

The clearest mismatch is fuel. The 1992 Principles say reactors should use only highly enriched uranium, reflecting the designs of the time. NSPM-20 treats low-enriched uranium as the lower-risk choice for non-proliferation reasons and pushes any other fuel into the presidential tier. A US reactor built on low-enriched fuel would be following domestic policy while departing from the letter of Principle 3. Because the Principles are not binding, that is not a breach of international law. But it shows how far the text has aged.

The second mismatch is propulsion. The Principles cover non-propulsive power. Nuclear electric and nuclear thermal propulsion, both named in current US plans, fall outside their stated scope.

The third is the lunar surface. The Principles are written around orbits and interplanetary trajectories, with re-entry as the main hazard. A reactor operating for years on the Moon raises different questions: siting, safety zones near other States’ activities, end-of-life disposal and contamination. Article IX of the Outer Space Treaty applies, but nothing in the 1992 text was designed for it.

The 2025 to 2026 push

Executive Order 14369, “Ensuring American Space Superiority”, signed on 18 December 2025, sets a goal of “deploying nuclear reactors on the Moon and in orbit, including a lunar surface reactor ready for launch by 2030.” On 14 April 2026 the White House Office of Science and Technology Policy issued memorandum NSTM-3 establishing the National Initiative for American Space Nuclear Power. It refers to “safely deploying nuclear reactors in orbit as early as 2028 and on the Moon as early as 2030”, directs NASA to start a programme for a mid-power reactor providing at least 20 kWe, and directs the Department of War to pursue a mid-power in-space reactor by 2031, pending funding. It also asks the Department of Energy to assess whether industry could produce up to four space reactors within five years, and to look for efficiencies in safety analysis and launch approval.

That last instruction is the one lawyers should notice. Launch approval under NSPM-20 is a deliberately layered process. Faster timelines will put pressure on it, and on the NEPA reviews that sit alongside it.

What I would watch

The first is the first reactor launch approval under NSPM-20. If SR-1 Freedom, NASA’s planned nuclear electric propulsion mission, keeps to its 2028 date, its tier, its safety analysis and the review board’s Safety Evaluation Report will set the template for the reactors that follow.

The second is commercial involvement. NSTM-3 expects private vendors to design reactors, and NSPM-20 makes the FAA the licensing authority for commercial launches. A privately launched reactor would test AC 450.45-1 in practice.

The third is the UN. As lunar plans multiply, I would expect pressure in the Committee on the Peaceful Uses of Outer Space to revisit the 1992 Principles, in particular the fuel rule and the absence of surface reactors, though agreement by consensus will be slow.

The fourth is liability on the Moon. If a reactor fails on the lunar surface and harms another State’s equipment nearby, the claim would most likely be judged under the Liability Convention’s fault standard for damage away from the Earth. That is a scenario the Convention’s drafters did not have in mind.