What an EMP Would Actually Do: The Evidence, Not the Sales Pitch

Electromagnetic pulse is the most heavily marketed fear in preparedness, and almost every page you'll find about it is attached to something for sale. That's unfortunate, because the underlying subject is real, well documented, and considerably more interesting than the version in the sales videos.

There is declassified test data. There is a congressionally mandated commission that ran physical experiments. There is a historical event from 1859 that we can measure against. This page is what that evidence shows.

An EMP is three different things

The single biggest source of confusion is that "EMP" describes one event with three distinct components, each affecting different equipment on entirely different timescales. Most popular writing collapses them, which is why the conclusions are usually wrong.

E1 is the one popular culture focuses on. It's fast enough that a surge protector cannot switch in time — the pulse is over before the device knows it started. Protection means a continuous conductive enclosure, or nothing.

E2 is the least discussed because it's the least novel. Infrastructure already handles lightning; E2 resembles it closely enough that existing protection mostly works. The concern is that E2 arrives immediately after E1 may have damaged the protection itself.

E3 is the component that actually threatens civilisation, and it's the one nobody makes videos about. It's slow, it couples into long conductors, and it damages the large power transformers that the grid depends on. It's also the component a geomagnetic solar storm produces — meaning the sun can do this without anyone launching anything.

What actually happened in 1962

On 9 July 1962 the United States detonated a 1.4-megaton device at roughly 400 km altitude over Johnston Atoll in the Pacific. The test was called Starfish Prime.

Roughly 1,400 km away in Honolulu, streetlights failed, burglar alarms went off, and a microwave link was disrupted. Several satellites in orbit were damaged over subsequent months by the artificial radiation belt the test created.

That's the strongest empirical datapoint available for a high-altitude nuclear EMP, and it cuts both ways. Real effects occurred at enormous distance — that's not nothing. But 1962 Honolulu was not reduced to a pre-industrial state, and the effects were localised to specific vulnerable systems rather than universal.

The honest caveat is that 1962 electronics were vacuum tubes and discrete transistors, which are dramatically more robust than modern integrated circuits. Contemporary equipment is far denser, runs at lower voltages, and is correspondingly more fragile. The Starfish data tells you an EMP has real reach; it does not tell you what happens to a 2026 device.

What actually happened in 1859

The Carrington Event, on 1–2 September 1859, was a solar coronal mass ejection that struck Earth directly. Telegraph systems across Europe and North America failed. Operators reported shocks, some telegraph paper caught fire, and in a few cases operators were able to send messages with the batteries disconnected, running on induced current alone. Aurora were visible close to the equator.

This is the E3 mechanism, from the sun rather than a weapon, and it's the reason the scenario is taken seriously by grid operators rather than only by preparedness marketers. The telegraph network was the only extensive electrical infrastructure that existed in 1859. There is a great deal more of it now.

The 1989 confirmation came on 13 March, when a geomagnetic storm collapsed the Hydro-Québec grid in roughly ninety seconds, leaving about six million people without power for around nine hours. That's a real, modern, documented grid failure caused by space weather — and it's a far better basis for planning than any hypothetical.

The car question, settled

The most persistent EMP claim is that every vehicle stops permanently. The EMP Commission tested this directly, and the results are in its published report.

Thirty-seven cars from model years 1986 to 2002 were exposed in an EMP simulator at progressively increasing field strengths. The most serious effect on running vehicles was that three cars' engines stopped at field strengths around 30 kV/m or above — vehicles that in a real exposure would coast to a halt and require a restart. About 25 cars showed malfunctions considered a nuisance, such as blinking dashboard lights, requiring no driver intervention at all. Below 25 kV/m, roughly 10% or fewer experienced serious effects like engine stall.

Not one was permanently disabled.

Two honest caveats. The testing stopped as soon as a vehicle showed any dysfunction, in order to avoid destroying it — so the data establishes thresholds for onset rather than for permanent damage. And no vehicle newer than 2002 was tested, while cars have become substantially more electronically integrated since. Hybrid and electric vehicles were never assessed at all.

But the finding that matters stands: the image of universally dead vehicles is not supported by the only physical testing anyone has published. The realistic scenario is a minority of stalls — which, on a highway at speed, is its own serious problem.

The thing that should actually worry you

Not your car. Not your phone. Large power transformers.

These are the units that step voltage up and down across the transmission network. They are custom-built to order, weigh hundreds of tons, cost millions, and have lead times measured in many months to years. A significant share of the supply is manufactured outside the United States. There is no meaningful spare inventory, because holding spares for bespoke multi-million-dollar equipment is not economic.

E3 and severe geomagnetic storms damage these through geomagnetically induced currents. If a substantial number failed simultaneously, the constraint on recovery would not be engineering knowledge or willingness — it would be manufacturing capacity and shipping.

This is the genuine tail risk in the whole subject, and it's the part the marketing skips, presumably because there's no consumer product that addresses it. Regulators do take it seriously: NERC's TPL-007 standard requires transmission operators to assess and plan for geomagnetic disturbance events.

What a Faraday cage does and doesn't do

A Faraday cage is a continuous conductive enclosure. Electromagnetic energy induces currents in the shell, which distribute around the outside and leave the interior largely field-free.

What matters: continuity of the conductive surface, and the contents not touching it. A metal container with a well-seated lid, with the contents insulated from the walls by cardboard or plastic, is the standard practical implementation. Gaps and seams are the failure mode — a cage is only as good as its worst opening.

What doesn't matter as much as people think: grounding. For E1 protection, an ungrounded enclosure works. Grounding is about safety and about E3 coupling in long conductors, not about shielding a box of electronics.

What it can't do: protect anything plugged in, anything with an antenna outside the enclosure, or anything you need to actually use during the event. A radio in a sealed can is protected and useless until you take it out.

The reasonable version of this is a spare radio, a spare charge controller or inverter, backup drives, and copies of essential documents — sealed, tested, and left alone. Not your working equipment.

What's actually worth doing

Everything that helps in an EMP scenario also helps in the vastly more likely scenarios: a winter storm, a hurricane, a substation failure, or an ordinary multi-day outage. That's the useful frame, because it means preparation is justified by events that happen regularly rather than by one that may never happen.

Water storage, food that needs no refrigeration, a heat source independent of the grid, a way to charge a phone, and cash. A spare radio in a sealed container is cheap and reasonable. A five-figure hardened bunker to defend against a scenario whose vehicle predictions are already contradicted by the only published testing is not.

The short version

An EMP is a real physical phenomenon with declassified test data behind it. Its most-marketed effect — every car dead forever — is contradicted by the only vehicle testing anyone has published. Its most serious effect, damage to large transformers with multi-year replacement lead times, is barely marketed at all, because nobody can sell you a transformer.

The sun can produce the dangerous component without any weapon involved, it has done so twice in recorded history at scales worth noting, and grid operators plan for it under a formal reliability standard.

Prepare for the outage, not the movie.

Sources

• Report of the Commission to Assess the Threat to the United States from Electromagnetic Pulse (EMP) Attack — vehicle testing results, p. 115
• Starfish Prime test data, Defense Threat Reduction Agency historical records
• Hydro-Québec March 1989 geomagnetic storm event reports
• NERC Reliability Standard TPL-007, geomagnetic disturbance planning

Last reviewed: July 2026