Radioactive Relics: The Missing RTGs

by Brian Dunning
February 10, 2026

Sponsor an episode!

The Arctic coast of Russia is about as remote and unforgiving a place as there is. Go east from Norway and you’ll follow 24,000 kilometers of coastline with countless inlets and peninsulas — from the Barents Sea to the Kara Sea to the Laptev Sea to the East Siberian Sea to the Chukchi Sea, until you reach the Bering Strait, beyond which is Alaska. Along that whole stretch are fewer settlements than you can count on the fingers of one hand. And yet shipping traffic has had to navigate the Northern Sea Route for a century, hauling millions of tons of cargo from one end to the other of the world’s largest country — most of it long before the invention of satellite navigation. Shore-based radio beacons were the primary navigational aid, which raised a problem: Without even a single road along the coast, let alone a power grid, how could the radio beacons be powered?

By the late 1980s, the Soviet Union installed over 1,000 radio beacons and lighthouses along their Arctic coastline, all powered by RTGs (radioisotope thermoelectric generators). Most of these were the Beta-M model which contained 280 grams of radioactive Strontium-90. Heat from the radioactive decay would flow through a thermocouple, producing a steady 10 watts of constant electrical power for some 10-20 years. The whole device is about the size of a barrel and is mostly big steel cooling vanes. If you didn’t know what it was, it would be pretty hard to guess.

Probably about 1,000 other Beta-M RTGs were manufactured for use in other remote parts of the country, but Soviet records were poorly kept and it’s estimated that perhaps several hundred remain abandoned throughout the Russian wilderness. They powered things like weather stations and radio repeaters, and are now — well — who knows?

Well some people know, because many of these devices have been vandalized, outright stolen, or even just stumbled upon. In Georgia in 2001, three woodcutters came upon the unshielded cores from two of these RTGs — meaning someone had already gone to a great deal of difficulty to disassemble them, a task which probably required cutting torches. The Strontium cores themselves were ordinary looking bars of metal, only 10 x 15 cm, though surprisingly heavy. The woodcutters found them sitting on the ground which was steaming, and all the snow had melted away around them. They figured they’d make a fine substitute for a campfire, so gathered around and settled down for the night. Within just a few hours all three men were vomiting profusely, and a couple weeks later they checked into the hospital and were diagnosed with acute radiation poisoning. Long story short, one died and another was hospitalized at a special clinic in Paris for more than 18 months. This event became known as the Lia radiological accident.

What of the unknown scrap metal thief who cut apart the RTGs? Those cores found by the woodcutters had been encased within one metric ton of 10 cm thick steel radiation shield; cutting them open was no small task at all. He (or they) probably never made it to a hospital — but we’ll likely never know.

But this was hardly the only such case; leave thousands of radioactive cores around a country unguarded, and you’ve got to expect some trouble. Here are other cases:

• In 1999, an unshielded core was found sitting at a bus stop near Leningrad. No casualties are known, but whoever handled it undoubtedly had very serious exposure.
  
  
• The same year, reindeer herders ran over an RTG with their truck in Chukotka, breaking it open. Five years later, the area maintained its status as an acute radiological hazard, still not cleaned up.
  
  
• In 2000, four idiots tried to remove the core from a lighthouse in Kamchatka planning to heat their apartment with it. After exposing the core they abandoned their project; no word on their condition.
  
  
• The same year, a group of naval personnel tried to transport an RTG to a lighthouse, but found it too heavy and abandoned it in the town of Plastun, creating a prolonged exposure hazard for the residents. The Russian Navy cleaned it up… 18 months later.
  
  
• In 2001, four people tried to disassemble an RTG lighthouse for its scrap metal in Murmansk; all four ended up in the hospital. Officials subsequently found three more lighthouses in the area similarly scavenged.
  
  
• In 2003, a core was recovered from the seafloor in the Gulf of Finland where thieves had dumped it after stealing the steel shielding. Two more RTGs were subsequently found completely disassembled in Kola Fjord, including the cores, and also missing the steel shielding. No word on the thieves in any of these cases, but they would have suffered very serious exposure.

A number of reports have been published by sources such as the Institute of Nuclear Material Management in 2016, a Russian report made to the International Atomic Energy Agency in 2013, and the Norwegian Radiation Protection Authority in 2005 and 2009. These reports document numerous additional cases where RTG-powered lighthouse and navigation beacons are simply gone, or are not found where they were supposed to have been installed, or (very often) damaged and stripped of metal and wiring. And moreover, the reports emphasize that all such reporting is incomplete and additional unknown incidents have no doubt occurred.

The United States also deployed RTGs around the world during that same era (and just to be clear, we’re limiting our discussion today to terrestrial RTGs, not to RTGs used aboard spacecraft and Mars rovers). Most visibly, ten such devices were built in 1973 to power seismic monitors and communication equipment at the Burnt Mountain Seismic Observatory in Alaska. This facility was created to detect seismic vibrations that would have been generated by any Soviet nuclear tests in Siberia, which were banned by treaty. These RTGs were generally similar to the Soviet types in their basic design and type of fuel used, and functioned until 2000 when they were replaced in order to meet increasing power requirements. The facility is now powered by a hybrid system incorporating batteries, solar panels, and wind generators.

Burnt Mountain, like other American RTG sites, was operated by the US military, with strict security and maintenance routines that prevented any known instances of thefts or vandalism. We didn’t have nearly as many RTGs as the Soviets, but we still had an estimated 100-150 deployed over the years, beginning in the 1960s and — so far as I could find — all taken out of service by 2000.

Most of our RTGs, such as the Sentinel series, were Strontium-90 powered, like the Soviets’ Beta-M. But a few, such as the SNAP series, used Plutonium-238. These were far more expensive, but had the following benefits:

• Plutonium-238 is much safer than Strontium-90. While Strontium emits beta particles, high energy electrons which can penetrate deeply, Plutonium emits alpha particles. These are basically Helium nuclei, big and heavy and very damaging, but only over a very short range, and easily stopped by a piece of paper or your skin’s outer layer of dead cells. You have to actually ingest Plutonium-238 to be harmed by it.
  
  
• Because of alpha particles’ short range, Plutonium RTGs require far less shielding, saving (literally) tons of weight. This is why they are still used today on spacecraft.
  
  
• They output more power and last three times longer, as Plutonium-238 has a half-life of 88 years compared to 29 years for Strontium-90.

So SNAP RTGs were used whenever applications required those benefits. Most didn’t:

• In Alaska, a SNAP-19 and a series of Sentinel RTGs were deployed at Fairway Rock Station on Rock Island in the Bering Sea in 1966 and 1981 to power oceanographic sensors, and in 1977 to power a string of five UHF relay stations to support air traffic control communications through Lake Clark Pass.
  
  
• Off the coast of California, the Navy placed an acoustic range transmitter powered by a Sentinel-25 2200’ underwater atop the San Juan Seamount 180 miles off San Diego in 1969; and in 1970, set up a Sentinel-8 powered meteorological data collection system on San Miguel Island. In 1974, the SEACON II experiment 2900’ feet underwater in the Santa Monica Basin was powered by a SNAP-21 RTG, recording the response of a large structure to changing ocean currents.
  
  
• In the Caribbean, a US Navy facility on the island of Eleuthera set up a Sentinel-100 in 1974, largely for experimental purposes. In 1976 they powered two experimental ocean buoys with RTGs: the BEAR buoy with a Sentinel-25, and the SEA ROBIN IV buoy with a SNAP-21 which almost became a problem. The buoy broke loose from its mooring and it was 19 days before the Navy was able to find and recover it with its RTG intact.
  
  
• In 1970, the Naval Avionics Center in Indianapolis set up a SNAP-7 for testing purposes.
  
  
• Off the coast of Panama City Beach, Florida in 1973, the Navy set up some ocean wave gauges powered by a Sentinel-25 on a drilling platform.
  
  
• Antarctica was (literally) a hotbed of RTG activity. Beginning in 1976, a whole series of Sentinel-25 powered PAWS (Polar Automated Weather Stations) were set up at Marble Point and also at Minna Bluff. Even out on the Ross Ice Shelf, the National Science Foundation and University of Alaska set up the SNAP-21 powered Windless Bight Infrasonic Observatory.

It’s worth mentioning that the Soviet Union also went RTG-crazy in Antarctica, setting up some 800 (!!) of them all around the continent. All of these were safely removed in a joint recovery operation with the United States beginning in the early 2000s.

But all of those US deployments are just the ones among those that have been publicly disclosed. In 1978 the US Navy Superbattery was demonstrated — a tiny RTG only about the size of a handheld flashlight and powered by Plutonium-238. Its stated intended purpose was to power undersea surveillance systems collecting data from undersea cables, but it was also stated that “numerous potential applications of this power source are anticipated.” It is probably safe to assume the Superbatteries, and their technological successors, found much such work; and are possibly even doing so today.

However, the Americans did not quite manage to maintain possession of 100% of their terrestrial RTGs; more like 99%. For it turns out one was lost and (so far) never recovered. One. Slightly better than the Soviets’ record of unknown hundreds remaining unaccounted for. It was to be installed atop India’s Nanda Devi mountain in the Himalayas to power a radio listening station spying on China’s nuclear missile program, but bad weather cut its installation short and it had to be left up there for months. When another crew returned, the RTG had disappeared for good — lost to a landslide or an avalanche.

Among many in India, this is quite the scandal. There are all kinds of pseudoscientific claims about what damage it’s doing up there. There are claims that it’s melting the glaciers and causing floods, even that it’s the true cause of storms and other crazy weather, and that it’s poisoning the Ganges with deadly radiation that could threaten millions. However, the possibility of negative health impacts are considered virtually nonexistent by experts. Even if one of the unit’s six fuel capsules ever did make it to a river, each is encased inside a tantalum liner inside a Haynes-25 alloy casing, all electron-beam welded into a solid metal bar. Both tantalum and Haynes-25 are used in applications such as jet engine turbine blades and afterburners due to their strength, durability, almost total immunity to corrosion, and temperature resistance. No radiation escapes them at all; only heat. Even if it did, alpha radiation has to be ingested to be harmful.

So the bottom line is yes, there are indeed RTGs out there in the world, posing a very substantial risk, and actively doing harm today. It is the Strontium-90 RTGs we need to be concerned about; and unless Russia makes some very deep systemic changes to practically everything about how they do anything, we’re going to continue to need to worry about them for a long time.

Special thanks to Chip Taylor, who provided voluminous background information to help get me started.

By Brian Dunning
Follow @BrianDunning