EMP and the Power Grid: Separating Facts From Fiction

Few grid threats attract as much hype as the electromagnetic pulse. Novels, films and online videos describe a single blast that ends modern life overnight. The real picture is less apocalyptic. EMP is a genuine, studied risk with its own executive order, but the best public evidence points to serious regional damage, not the end of civilization. This guide separates what is known from what is claimed.

Key facts

• A nuclear EMP has three components: E1 (under a microsecond), E2 (up to about 10 milliseconds) and E3 (seconds to minutes), each threatening different equipment (CISA).
• The 1962 Starfish Prime test, a 1.4-megaton burst 400 km above the Pacific, is the main real-world U.S. data point (LLNL). About 30 strings of streetlights failed on Oahu, roughly 1,445 km away.
• EPRI's 2019 study found a single high-altitude burst could cause a regional blackout but not a nationwide grid failure, with 3 to 21 large power transformers at risk.
• The widely repeated claim that 90% of Americans would die traces to a 2008 congressional exchange about a novel.
• Executive Order 13865, signed March 26, 2019, set federal EMP resilience policy and treats geomagnetic storms as a natural type of EMP.
• CISA says a HEMP attack by an adversary with basic nuclear weapons and missile capabilities "may be disruptive on a regional scale" but is unlikely to cause catastrophic damage to the grid on a continental scale (CISA).

What an EMP is

An electromagnetic pulse is a burst of electromagnetic energy. The scenario that worries planners most is a high-altitude electromagnetic pulse (HEMP): a nuclear weapon detonated above the atmosphere, sending electromagnetic energy toward the surface over a wide area (EEI summary of EPRI). CISA says the footprint can range from regional to continental depending on the height of the burst.

The three parts of a nuclear EMP

Component	Duration (CISA)	Behaves like	Equipment most at risk
E1	Less than 1 microsecond	A very fast, high-peak pulse	Electronics, control systems, telecommunications, protective relays
E2	Less than 10 milliseconds	Nearby lightning	Power lines and towers, electronics, transformers
E3	About 10 seconds (blast) to 1–2 minutes (heave)	A severe geomagnetic storm, possibly with about 3 times the field	Transformers, protective relays, long transmission and communication lines

Source: CISA EMP Protection and Resilience Guidelines (2019), adapted from DOE.

Nuclear vs. non-nuclear EMP

Not every EMP involves a nuclear weapon. CISA lists three sources: high-altitude nuclear detonations, specialized conventional munitions, and non-nuclear directed-energy devices. A radio-frequency weapon aimed at a single data center or substation could cause "significant localized or regional" disruption, according to CISA's guidelines, but nothing like the multistate footprint of a HEMP.

Starfish Prime: the one real-world data point

On July 9, 1962 (about 11 p.m. on July 8, Hawaii time), the United States detonated a 1.4-megaton warhead 400 km above Johnston Atoll in the Pacific.

The most cited damage is on Oahu, about 1,445 km away. Glasstone and Dolan's The Effects of Nuclear Weapons (DoD and DOE, 1977), a widely used government reference, called the simultaneous failure of 30 strings of streetlights there one of the best authenticated cases of EMP damage. Sandia physicist Charles Vittitoe reviewed the incident and concluded the damage was EMP-generated, in a way consistent with only some strings failing (Sandia; OSTI abstract). His report also notes another researcher's point that 30 strings were a small percentage of the system and "not indicative of mass destruction of electrical systems."

The lesson cuts both ways: EMP effects are real at long distances, but the observed damage was limited, and 1962 electronics were very different from today's.

The EMP Commission and its contested claims

Congress created the Commission to Assess the Threat to the United States from EMP Attack in 2000 and reauthorized it in 2006 (Senate Republican Policy Committee). In its April 2008 report on critical national infrastructures, the Commission, chaired by William Graham, warned that restoring even a diminished but workable level of electric service "could well take many weeks, with some probability of it taking months and perhaps more than a year at some or many locations."

Where the "90% would die" figure comes from

At a July 10, 2008 House Armed Services Committee hearing, Rep. Roscoe Bartlett described the plot of the novel One Second After, in which 90 percent of the population is dead after a year, and asked whether it was "a realistic assessment." Graham answered: "We think that is in the correct range."

That exchange is the documented origin of a number that has been widely repeated since. Two points are fair to both sides:

• Critics' view. The number began as a fictional scenario. We searched the Commission's 2008 infrastructure report and found no 90 percent mortality estimate or casualty model behind it.
• Supporters' view. The Commission did warn that losing power and dependent services across large regions for months would be catastrophic, and that society "as we know it" could not exist within those regions.

In short, 90 percent is an expert's judgment about an extreme, nationwide, long-duration scenario, not a modeled result, and later studies dispute that scenario.

What the Commission's own car tests showed

The same report undercuts a popular myth that every car would die. The Commission exposed 37 cars, model years 1986 to 2002, to simulated EMP up to about 50 kV/m. Cars that were turned off showed no effects. Three running cars stalled at about 30 kV/m or above, one needed dashboard repairs, 25 showed only nuisance effects, and eight showed none. The Commission expected "few automobile effects" below 25 kV/m and serious effects in about 10 percent or more of cars at higher levels, mainly a risk of crashes.

What EPRI's 2019 study found

The Electric Power Research Institute launched a research project in 2016 with DOE, Lawrence Livermore, Sandia and Los Alamos national laboratories, the Defense Threat Reduction Agency and industry, and released its final report on April 30, 2019. According to EPRI and the Edison Electric Institute's summary of the report:

• A moderate E1 pulse could disrupt or damage about 5% of digital protective relays at transmission line terminals across an interconnection, and a severe one about 15%. That is unlikely to cause an immediate interconnection-wide blackout.
• Relays were generally resilient to the pulse itself but vulnerable to surges induced on their control and communication cables.
• E2 is not expected to damage the transmission system.
• E3 could cause a regional blackout across multiple states, but immediate, widespread transformer damage is not expected. Only 3 to 21 large power transformers were found to be at risk.

"Our findings do not support the belief that HEMPs can cause nationwide blackouts lasting for years," said EPRI's Randy Horton. EPRI identified mitigations including shielded and grounded cables, surge protection, fiber-optic communications and spare relays and transformers.

The critics

Not everyone accepts EPRI's conclusions. An August 2019 rebuttal from the Electromagnetic Defense Task Force argued that EPRI used an E3 field of about 35 V/km, while a declassified analysis of 1962 Soviet tests supports about 85 V/km. It also said EPRI chose less damaging burst heights, did not model multiple bursts, and studied only the transmission system, leaving out generation, distribution, communications and pipelines. CISA's own guidelines tell planners to use 85 V/km for the E3 threat to the continental U.S.

The disagreement is mostly about assumptions: field strength, burst height, number of weapons and scope. Both sides agree EMP is a real risk.

Executive Order 13865

On March 26, 2019, Executive Order 13865 made it U.S. policy to prepare for EMPs through whole-of-government planning and private-sector engagement. It defines a geomagnetic disturbance as "a type of natural EMP." Among its tasks:

• DHS was to identify priority critical infrastructure within 90 days and assess which is most vulnerable to EMPs.
• DOE was to develop or update quantitative EMP benchmarks for grid owners and operators within one year.
• DoD was to run a pilot test hardening a strategic military installation.
• Interior was to complete a magnetotelluric survey of the contiguous U.S. within four years, mapping ground conductivity, which affects how strong induced currents become.

The order followed a 2016 GAO report finding that DHS had not fully assessed electromagnetic risks; GAO now lists those recommendations as implemented.

EMP vs. solar storms: where E3 and GMD overlap

CISA describes E3 as "similar to a larger version of solar geomagnetic disturbance," and says a nuclear E3 pulse can be significantly more intense. The main differences:

• Warning. CISA says a solar storm typically gives 18 to 72 hours of strategic warning and 20 to 45 minutes of tactical warning. A nuclear EMP might give none.
• Duration. E3 lasts seconds to minutes; geomagnetic storm effects can last hours.
• E1. Solar storms have no equivalent of the fast E1 pulse.

That overlap is useful. EPRI lists blocking geomagnetically induced currents and keeping spare transformers among its E3 mitigations, the same kinds of tools the industry uses against solar storms. Read more in our guide to solar storms and the power grid.

Fact vs. fiction

Claim	What the evidence says
"One EMP would fry every device in America."	HEMP footprint is regional to continental depending on burst height; E1 effects vary widely by equipment.
"Every car would stop."	In Commission tests, 3 of 37 running cars stalled; none that were off were affected.
"Every big transformer would burn out."	EPRI found 3 to 21 at risk; critics argue stronger E3 fields are possible.
"90% of Americans would die."	An expert's endorsement of a novel's scenario, not a published model.
"EMP is pure fiction."	Starfish Prime caused real damage 1,445 km away, and federal policy treats EMP as a real risk.

Realistic preparedness

CISA calls extreme EMP incidents "hard problems" and "low probability/high consequence" scenarios. It also notes that adversaries with highly developed nuclear capabilities "might cause widespread harm to U.S. infrastructure" with complex HEMP attacks during an escalating international conflict, a scenario in which an EMP would be one problem among many.

For households, preparing for EMP looks like preparing for a long regional outage of any cause:

• Water and food for at least several days, ideally longer. Our grid-down preparedness checklist has quantities, and our guide to how long food lasts covers the fridge and freezer.
• Medicines and medical devices. Talk to your doctor about backup plans.
• Communication without the grid. Spare phone batteries or a hand-crank or solar charger, a corded landline handset if you have a traditional line, and a paper contact sheet with an out-of-town contact (National Weather Service).
• Safe backup power. If you use a generator, follow generator safety rules.
• Skepticism. Be wary of products marketed with fear-based EMP claims. Basic outage preparation covers the most likely problems.

Our Grid Threat Monitor tracks the more common threats, and our guide to cyberattacks on the grid covers another scenario that often gets exaggerated.

The bottom line

EMP is a real, studied risk, not a myth. But the most detailed public engineering study found that a single high-altitude burst would most likely cause a regional blackout with limited transformer damage, not a years-long national collapse. Critics argue the worst case could be worse, and that debate continues. Some of the most dramatic claims, including that 90% of Americans would die and that all cars would stop, are not supported by published evidence. Prepare for a long outage, and leave the panic to fiction.

PowerOutageTracker.com. "EMP and the Power Grid: Separating Facts From Fiction." Accessed November 2, 2026. https://poweroutagetracker.com/guides/emp-and-the-power-grid-facts-vs-fiction