Nuclear programs: what the government confirmed

Between 1945 and 1962, approximately 225,000 members of the U.S. Armed Forces participated in hundreds of nuclear weapons tests. They became known as Atomic Veterans. The exercises at the Nevada Test Site were designated Operation Desert Rock and ran across six exercises from 1951 through 1957, with total Department of Defense participation exceeding 68,000 personnel across those exercises alone. Additional soldiers observed tests in the Pacific.

The stated purpose was to train troops for operations on a nuclear battlefield and to study the effects of nuclear detonations on military equipment and personnel. Soldiers were placed in trenches within miles of detonation points. Some were ordered to advance toward the blast site within minutes of detonation. Observers described watching the bones of their hands through closed eyelids, the detonation bright enough to penetrate both skin and eyes. They were sworn to secrecy. The classification of their participation extended to their own medical records. Many could not tell their doctors what they had been exposed to or where they had served.

In 1982, a former Army medic named Van Brandon gave a statement to federal investigators disclosing that he had followed orders to prepare two sets of radiation records at four nuclear tests in 1956 and 1957 at Yucca Flats, Nevada, and had observed falsified documents at a fifth test in November 1955. One set, which he called the hot set, contained the true radiation readings. The second set, submitted as the official record, showed the soldiers had received less than the maximum permitted level of exposure. Brandon said when he left the Army in 1961 he was warned that if he disclosed what he had seen, consequences would follow. He disclosed it twenty years later anyway.

In 1990, Congress passed the Radiation Exposure Compensation Act, providing a one-time payment of $75,000 to qualifying atomic veterans. The documentation requirements for the claim process created a secondary problem: a 1973 fire at the National Personnel Records Center in St. Louis destroyed approximately 80 percent of Army personnel records for soldiers discharged between 1912 and 1960. For many atomic veterans, the records needed to prove their service no longer existed. The gap between what the government knew about exposure levels and what it put in official records, combined with the destruction of the records themselves, has never been formally accounted for in any released document.

The accident at Three Mile Island Unit 2 began at 4:00 a.m. on March 28, 1979. A cooling system malfunction triggered a partial core meltdown, the most serious accident in U.S. commercial nuclear power history. The public was informed that the situation was under control. Governor Richard Thornburgh recommended on March 30 that pregnant women and preschool children within five miles of the plant evacuate as a precaution. The NRC and plant officials debated for days whether to recommend broader evacuation and what the public should be told.

The NRC commissioned two major investigations. NUREG-0600, the Office of Inspection and Enforcement investigation, was produced internally. NUREG/CR-1250, the Rogovin Report, was an independent investigation commissioned by the NRC and produced by attorney Mitchell Rogovin and George T. Frampton. The Rogovin Report runs to multiple volumes and is among the most detailed official investigations of any nuclear accident in U.S. history.

What the NRC's own record establishes: uncertainty within the power plant meant that information provided to government agencies was fragmented, sometimes contradictory, and generally understated the severity of the accident in communications with the public. The NRC was receiving incomplete and sometimes inaccurate information from plant operators during the most critical hours. Those partial assessments were then communicated to state officials and to the public as the current understanding of the situation. The reports document the gap between what was known inside the plant and what was communicated externally, and attribute it to the fragmented and rapidly changing information environment rather than deliberate concealment. Whether that distinction holds across all communications during the five-day crisis is a question the Rogovin Report itself declines to fully resolve.

In the spring of 1964, Lawrence Livermore National Laboratory began a classified experiment to answer a question that had become urgent as more nations developed nuclear ambitions: could a small group of scientists, working without access to classified information, design a credible nuclear weapon from publicly available sources alone?

The laboratory selected two recent PhD physicists, David N. Pipkorn of the University of Illinois and David A. Dobson of the University of California, Berkeley. They were given access to a Livermore laboratory and its unclassified resources but were explicitly prohibited from accessing classified nuclear weapons information. When Pipkorn later moved to full-time Livermore work, Robert W. Selden of the University of Wisconsin joined. All three worked part-time on the project alongside other duties.

The experiment concluded in the spring of 1967, three years after it began. Dobson and Selden had produced a design for a plutonium implosion weapon with the potential for high explosive yield. One participant described the project as truly a do-it-yourself project. A State Department internal announcement distributed to government agencies for a forthcoming briefing on the results stated that three young PhD physicists, working part-time, succeeded in achieving a workable nuclear weapons design in a period of about three years. The two physicists then traveled to brief the CIA, Los Alamos, the Atomic Energy Commission, the State Department, and other agencies on what they had found. They delivered their briefing in the format of a mock Huntley-Brinkley evening news report.

The conclusion of the Nth Country Experiment was that a government with nuclear weapons aspirations and limited resources could develop a credible weapon from open-source scientific literature. The experiment began before China conducted its first nuclear test, when the question of how many countries could independently develop nuclear weapons was the central proliferation concern of U.S. policy. The 1967 Summary Report on the Nth Country Experiment was released to the National Security Archive in the 1990s. A massively redacted companion document, the Postshot Activities report summarizing the classified briefings the physicists gave across the country, was newly declassified and published by the National Security Archive on January 23, 2025.

In 2001, former Sandia National Laboratories safety official William L. Stevens wrote a history of U.S. nuclear weapons safety efforts, focusing specifically on the decades-long project to mitigate the risk of accidental or unsanctioned detonation. The document was classified. In 2022, the National Security Archive obtained it through a Freedom of Information Act request. The Department of Energy had previously restricted its distribution and reclassified portions as Secret.

The history addresses the problem of sealed-pit nuclear devices, the standard warhead design in the U.S. stockpile since the late 1950s, and the possibility that severe environmental insults to such a weapon, including fires, crashes, or other damage, could trigger detonation. Stevens wrote that the public must be encouraged to realize that risks of an unintentional nuclear detonation cannot be zero and cannot ever be really known.

The document also addresses specific incidents and near-accidents involving U.S. nuclear weapons, including issues with the Short Range Attack Missile (SRAM) and Minuteman ICBM warheads. The Air Force's institutional resistance to safety modifications is documented, with Stevens noting that the Air Force resented the implications of what he called goofproof safety mechanisms, viewing them as a suggestion that operators could not be trusted. The tension between operational readiness and safety engineering in nuclear weapons programs is a recurring theme in the history, and the document was restricted from wide distribution precisely because that tension had not been fully resolved.

The United States nuclear stockpile number, the total count of warheads in the arsenal, was declassified by the Biden administration in 2021 for all years through September 2020. It stood at 3,750 warheads. Subsequent annual requests from the Federation of American Scientists to disclose stockpile numbers for 2021, 2022, and 2023 were all denied. The Trump administration has not restored stockpile transparency. The current size of the U.S. nuclear arsenal is classified.

That gap, between the historical record that has been opened and the current operational reality that has been closed again, is itself a primary source fact. What the government chooses not to release is as significant as what it releases. The nuclear programs documented on this page span from 1951 to the present. The most recent information, the current stockpile number, is the most unavailable.