Short, sourced answers on nuclear safety, waste, cost and radiation — written to be used in a real conversation, not to win an argument online.
Nuclear is among the safest energy sources ever measured — on par with wind and solar.
Counting every accident plus the air pollution each source causes, nuclear results in about 0.03 deaths per terawatt-hour of electricity — versus 24.6 for coal, 18.4 for oil, and 2.8 for natural gas. Modern plants layer redundant, passive safety systems behind metre-thick containment.
Chernobyl's fatal combination doesn't exist in a Western reactor.
Chernobyl was a 1986 Soviet RBMK: no containment building, and a core that produced more power as its coolant boiled away — a strongly positive void coefficient. Western light-water reactors are the opposite on both counts. Water is the moderator as well as the coolant, so losing it shuts the reaction down rather than accelerating it, and the whole thing sits inside a steel-and-concrete containment building. RBMKs still run in Russia, modified after the accident; no Western regulator would license the original design. Chernobyl's other legacy is the one worth naming: around 19,000 thyroid cancers among people who were children in 1986, caused by contaminated milk the Soviet authorities did not withdraw. Thyroid cancer is highly survivable — about 15 of those cases have proved fatal — and the failure was a government's, not a technology's.
Fukushima's radiation is linked to one death. The evacuation is linked to 2,313.
A once-in-a-millennium earthquake and tsunami overwhelmed a 1960s-era plant. Among the public, the UN's scientific committee (UNSCEAR) and the WHO found no detectable health effects from the radiation and expect none. One death has been officially attributed to it: a plant worker whose lung cancer Japan's health ministry recognised as radiation-related in 2018. Fukushima Prefecture's official count of 2,313 disaster-related deaths — people who died in the upheaval rather than in the wave — is attributed largely to the rushed evacuation, driven by fear more than by measured risk. The ~18,000 lives lost in the region were taken by the tsunami, not the reactors. Modern designs add passive cooling that prevents this kind of meltdown.
Modern plants are built to contain the worst — and the physics makes a Chernobyl-style runaway impossible.
Western reactors sit inside steel-and-concrete containment domes engineered to survive earthquakes and, for new designs, a direct hit from a large aircraft. The physics works in your favour twice over: as the fuel heats, it absorbs more neutrons and the reaction slows (the Doppler effect), and if the coolant boils away it takes the moderator with it, so the reaction stops rather than runs away. Newer designs add passive cooling — an AP1000 is licensed to keep its core safe for 72 hours with no power and nobody in the control room, using only gravity and natural circulation. Multiple independent barriers, 'defence in depth', are why nuclear's real-world safety record sits alongside wind and solar.
A single flight gives you more than a year spent living next door to a plant.
Everything is mildly radioactive — granite benchtops, brick walls, your own body. Living within a mile of a nuclear plant adds well under 1 millirem a year, against the roughly 300 millirem everyone receives from nature and about 4 millirem from a Sydney–London flight. Coal is the awkward comparison for coal: burning it concentrates the uranium and thorium already in the rock and sends some of it up the stack, so per unit of electricity a coal plant releases considerably more radioactivity into its surroundings than a nuclear plant does — because nuclear's stays inside the fuel.
All US nuclear waste ever produced would fit on a single football field.
It's solid ceramic pellets sealed in steel and concrete, and it doesn't glow green. Every bit of used fuel from 60+ years of US commercial reactors would cover one football field to a depth of under 10 yards. It is also the only industrial waste stream in the world where every gram is counted, tracked and paid for by whoever made it. What it lacks is a permanent home. Finland's Onkalo repository cleared its safety review in August 2026 and awaits a government licence — it would be the first anywhere. Sweden and Canada have chosen sites. The US has not: Yucca Mountain has been unfunded since 2010, so American fuel sits in dry casks on the plant sites that made it, which is safe but not final. Two US spent-fuel pools, at Indian Point, have leaked tritium into groundwater — detected, monitored, below drinking-water limits, and a reason to say 'well-contained' rather than 'never leaks'.
Dangerous for centuries; back to ore level in about 100,000 years — which is exactly why we bury it.
Two different questions get mixed up here, so it's worth separating them. The intense heat and radiation fade fast: after about 500 years, spent fuel has lost the great majority of its radioactivity, and the fission products responsible are largely gone within a thousand. But returning all the way down to the radioactivity of the uranium ore it was made from takes on the order of 100,000 years, because of the plutonium and minor actinides. That long tail is the whole reason deep geological repositories are engineered the way they are. Reprocessing changes the arithmetic. Strip out the plutonium and actinides and burn them as fuel — as France does today, and as 'fast' reactors would do far more completely — and what's left decays to ore level in roughly a thousand years instead. Turning a hundred-thousand-year problem into a thousand-year one is the strongest single argument for recycling used fuel rather than burying it whole.
Expensive to build once; cheap, stable power for decades afterwards.
First-of-a-kind plants are costly — Vogtle and Hinkley Point C both ran far over budget, and pretending otherwise helps nobody. But cost falls with repetition: France built 56 reactors on a handful of standardised designs in about fifteen years, and South Korea has kept its build costs among the world's lowest by doing the same thing continuously since the 1980s. Fuel is only a small slice of the price, so nuclear electricity barely moves when gas prices spike, and US regulators now licence reactors out to 80 years — a paid-off reactor is one of the cheapest sources of electricity there is. France draws about 68% of its electricity from nuclear and has one of the lowest-carbon grids of any large economy. Its household electricity is cheaper than Germany's or Italy's but not the cheapest in Europe — the Nordics, on hydro, are lower.
Usually 5–10 years — and far faster when a country builds the same design repeatedly.
Recent first-of-a-kind Western projects — Vogtle in the US, Hinkley Point C in the UK — ran years late and billions over. That's the fair criticism, and it's about first units, not about nuclear. The UAE's Barakah plant shows both halves of the story honestly. Its first reactor was meant to enter commercial operation in 2017 and did so in April 2021, four years late. The remaining three then followed at roughly one a year, finishing in September 2024 — about 5.6 GW in total, on a repeated design with a workforce that had already built one. The whole station took twelve years from first concrete to the last unit in service; the three that followed the first arrived within three and a half years of it. That second half is the argument: the learning curve is real, and it is steep. France's 56 reactors in about fifteen years were built the same way. Small modular reactors aim to push it further by building the reactor in a factory rather than pouring each one on site.
Cheap to generate — but firming a 24/7 grid is where the costs hide.
Per unit generated, solar and wind are genuinely cheap. The expense shows up at the system level: to run a grid on them alone you need vast overbuilding, long-distance transmission, and days of storage for the windless, cloudy stretches — and that last 10–20% gets very pricey. Nuclear delivers firm, around-the-clock power that fills exactly that gap, which is why the cheapest clean grids tend to use both.
They're intermittent; nuclear is firm, 24/7 power.
Solar and wind are great but only work when the sun shines and the wind blows. Nuclear provides reliable power around the clock, running above a 90% capacity factor. Germany is the cautionary tale, though it needs stating accurately: its power-sector emissions have fallen sharply and coal generation is now at its lowest in seventy years, so 'Germany's emissions went up' is out of date. The real cost was the counterfactual. Germany closed 20 GW of working, paid-for, zero-carbon reactors and spent the following two decades using renewables to replace them instead of to replace coal. Had it retired coal first, it would be at a French-level carbon intensity by now.
Yes — over its whole life cycle, nuclear emits about as little CO₂ as wind.
The 'but concrete and mining!' objection doesn't survive the numbers. Counting everything — building the plant, mining and enriching the fuel, 60–80 years of operation, and decommissioning — the IPCC puts nuclear at a median of about 12 grams of CO₂ per kWh. That's on par with onshore wind (11 g), lower than solar (48 g), and a small fraction of natural gas (490 g) or coal (820 g). Nuclear is one of the lowest-carbon ways to make electricity yet measured.
More than other thermal plants, in fact — and it's still the better trade.
This is the one where the honest answer is a concession. Nuclear does use more water per megawatt-hour than coal or gas, because it runs at lower steam temperatures and so converts a smaller share of its heat into electricity; the rest has to be carried away. A sceptic who brings this up is right, and pretending otherwise loses the argument. What that costs is smaller than it sounds. Most of the water is withdrawn and returned warmer, not consumed — the consumed fraction is a few per cent. Coastal plants use seawater and touch no fresh water at all, which is the obvious siting for Australia. Recirculating towers cut withdrawals by around 95%, and dry air cooling eliminates them at a small efficiency penalty, which is how reactors are being planned for inland and desert sites. And the comparison that matters is per unit of round-the-clock electricity across everything, not water alone: on land and on materials, nuclear is among the lightest-footprint sources there is.
The risk sits in enrichment and reprocessing plants, not in power reactors — and it's a political problem, not a physical one.
Commercial fuel is enriched to 3–5% uranium-235. A weapon needs around 90%, and anything above 20% is legally 'highly enriched' for a reason: most of the separative work is in the first stretch, so the machinery that takes you to 5% is the machinery that matters. That's why safeguards concentrate on enrichment and reprocessing facilities rather than on reactors. The honest version isn't that reactor material can never be used. The US demonstrated a device using reactor-grade plutonium in 1962, and India's 1974 test used plutonium from a research reactor Canada had supplied. What the record actually shows is that the two paths are separable and separated: over thirty countries run power reactors and don't build weapons, every weapons state built its arsenal through dedicated military facilities rather than its power programme, and the safeguards system exists precisely because the link is political. Australia, which already hosts the IAEA's regional safeguards experience and sells uranium under bilateral agreements, is about as well-placed on this as a country can be.
No — over a century of known reserves, and effectively limitless with recycling.
Identified uranium resources cover well over a century at today's use, and recycling spent fuel or using 'breeder' reactors stretches that to thousands of years. There's also uranium dissolved in seawater in near-limitless quantity. Fuel supply is not the constraint on nuclear power.
Smaller, factory-built reactors meant to be faster, cheaper, and deployable almost anywhere.
Instead of one giant custom-built plant, an SMR is a smaller reactor — usually under 300 MW — built on a production line and shipped to site. The pitch: factory manufacturing lowers cost and build time, the compact size suits smaller grids and industrial users, and many designs use passive safety that needs no operator action or outside power to stay safe. It is still a pitch. Two SMRs are operating — Russia's floating Akademik Lomonosov and China's HTR-PM — and several are under construction, but no Western SMR has been built and priced in a competitive market. The first-of-a-kind units now being costed come in around $4–6 billion, not the $1–3 billion often quoted, because the savings depend on repetition that hasn't happened yet. The technology is real, the economics are a forecast, and the difference is worth being straight about.