9 storm shelter myths that could get you killed

Bad tornado advice does not spread because people are careless. It spreads because it sounds right — a bridge looks like cover, the southwest corner sounds like physics, "FEMA-approved" sounds like a government stamp. The trouble is that every myth below has been tested by real events, and the results are on the record. We have pulled the primary sources for each: National Weather Service post-event analysis, NOAA National Severe Storms Laboratory fatality research, FEMA design guidance, and the ICC 500 test protocols. We do not sell shelters, so we have no reason to talk you into one or out of one.

Myth 1: A highway overpass is good shelter

This is the most dangerous myth in American severe weather, and it exists almost entirely because of one video. On April 26, 1991, a Kansas television crew filmed themselves surviving a tornado from beneath a Kansas Turnpike overpass. The footage ran nationwide. What the footage did not convey: that bridge had an unusual crawl space and exposed girder to hold onto, the tornado passed to the side rather than over it, and it was a comparatively weak tornado in open country.

Eight years later the myth was tested properly. On May 3, 1999, tornadoes crossed central Oklahoma interstates at seven locations. At three of those, an overpass was directly in the path — and there was a fatality at all three. National Weather Service forecasters in Norman documented the outcomes in detail:

• At the Shields Boulevard overpass in Moore, roughly a dozen people sheltered under the bridge. One woman was killed; her body was not recovered for a week because it lay beneath six to eight feet of debris. Nearly every survivor under that bridge suffered serious injuries from flying debris — compound fractures, amputated fingers and ears, impalement.
• That overpass had no girders or beams at all — smooth concrete overhead and smooth embankment below. Nothing to hold, nothing to hide behind.
• A third death occurred in rural Payne County where the tornado was only F2 intensity. You do not need a violent tornado to be killed under a bridge.

The physics runs the wrong way from intuition. Wind speed decreases as you approach the ground and reaches zero at the surface, so climbing up under the girders moves you into faster wind, not slower. Forcing air through the narrow gap beneath a bridge can accelerate it further. And the space beneath and just downstream of an overpass can become a debris deposition zone, which is exactly what happened at Shields Boulevard.

There is a second harm that rarely gets mentioned: cars stopping under an overpass block the highway, trapping other drivers inside the tornado's path with no remaining options. The NWS position is blunt — apart from getting out of the tornado's path entirely, there is no good option once you are in it, and the overpass is among the worst.

Myth 2: An interior closet or bathroom is basically a storm shelter

This one deserves a careful answer, because the interior-room rule is good advice and we do not want to talk anyone out of it. Moore, Oklahoma emergency management estimated more than 2,000 people were in the path of the May 3, 1999 tornado as it crossed the city. Five died. Only three of those were in well-built structures following standard guidance. One photographed interior room was the only part of a house left standing in a violent-damage zone, and seven people walked out of it with minor injuries. Interior rooms work, and they work far more often than the news coverage suggests.

What they do not do is provide what FEMA calls near-absolute protection. The distinction is engineering, not opinion:

The honest framing: an interior room is what you use because it is what you have, and it is dramatically better than an exterior wall, a window, a vehicle, or a bridge. A shelter is what you buy when you decide the EF4-and-above case matters to you. Our safety and standards page explains what P-361, ICC 500, and the NSSA seal each actually cover.

Myth 3: The southwest corner of the basement is the safest spot

This one is genuinely ancient — it traces to nineteenth-century advice reasoning that because most tornadoes travel southwest to northeast, debris would be thrown past the southwest corner rather than into it. It has been out of official guidance for decades, and it still gets repeated at family gatherings every spring.

It is wrong for two reasons. Debris inside a tornado's circulation does not travel in one tidy direction; it rotates, and it arrives from all sides. And the primary hazard in a basement is not the tornado's wind at all — it is the house above you collapsing into the basement, which can fill any corner. Current guidance is to get to the center of the lowest level, under a sturdy workbench or stairwell if one exists, away from exterior walls and windows, and specifically away from the floor space directly beneath heavy objects on the level above: refrigerators, water heaters, pianos, gun safes. Cover your head and neck.

Myth 4: Open your windows to equalize the pressure

Houses do not explode from pressure differential. They fail from wind loading and debris impact — a window breaks, the building pressurizes, the roof lifts, the walls lose their bracing. And an ordinary house is leaky enough that any pressure difference equalizes on its own through soffits, vents, and gaps you cannot see.

What makes this myth lethal is the time it costs. NOAA puts average tornado warning lead time at roughly 13 minutes. Walking a house opening windows can easily consume half of that, and it ends with you standing next to glass. Spend those minutes getting people, shoes, and a phone into the shelter instead.

Myth 5: "FEMA-approved" means the government tested this shelter

There is no such thing. FEMA does not approve, endorse, or certify shelters, products, or companies. It publishes design guidance: FEMA P-320 for homes and small businesses, FEMA P-361 for community safe rooms. Neither is a certification program, and neither is a building code.

The enforceable standard is ICC 500, the ANSI-accredited standard co-published by the International Code Council and the National Storm Shelter Association, which becomes law where a jurisdiction adopts it. Compliance is demonstrated through third-party testing and engineering documentation — much of the debris-impact testing in this industry is done at the Texas Tech University National Wind Institute.

Myth 6: Modern tie-downs make a manufactured home safe

Anchoring systems have genuinely improved, and the manufactured housing industry has argued for years that warnings singling out mobile homes are unfair. The fatality record does not support the argument. Harold Brooks of NOAA's National Severe Storms Laboratory went through Storm Data to test the claim directly:

The reason anchoring does not close the gap is that anchoring solves the wrong problem. Tie-downs resist uplift and sliding. They do nothing about a 2×4 traveling at highway speed through a wall assembly that is a fraction of the thickness of a site-built home's. That is why NOAA and FEMA guidance for manufactured-home residents is the same as it has been for thirty years: leave before the storm arrives for a community shelter or a sturdy building, and know in advance exactly where you are going and how long it takes to get there.

Myth 7: Rivers, hills, and downtowns break tornadoes up

Every region has its own version — the bluffs protect us, the river splits them, the heat island downtown pushes them around. There is no evidence for any of it. Tornadoes have been reported in all 50 states. Two frequently cited counterexamples: a violent tornado in July 1987 tracked 24 miles through Wyoming's Teton Wilderness, crossing the Continental Divide at over 10,000 feet; and a November 2005 tornado at Evansville, Indiana crossed the winding Ohio River three times. Dallas, Oklahoma City, St. Louis, Wichita Falls, Miami, and Salt Lake City have all taken direct hits.

This myth rarely kills on its own. It kills by causing delay — the few minutes spent deciding whether this one counts, when the warning already gave you the answer.

Myth 8: I can drive away from it, or drive to a public shelter when the warning hits

With an average warning lead time near 13 minutes, a plan that begins with driving is a plan that begins with traffic. Tornadoes do not move in straight lines, they are often rain-wrapped and invisible, and roads run where roads run, not where you need to go. On May 3, 1999 the highway itself became the trap: vehicles stopping at overpasses blocked the road and left everyone behind them with no options.

Driving is the right call in one specific situation — you are in a manufactured home or a vehicle, you have clear early warning well before the storm arrives, and you have a known destination a short, uncongested distance away. That is a decision made when the watch is issued, not when the warning sounds. If you are caught in the open with no building available, current guidance is to get out of the tornado's path if you can, and if you cannot, to lie flat in a low spot away from vehicles and cover your head — a last resort, not a plan.

Myth 9: Any concrete box in the ground is a storm shelter

Buyers reasonably assume the walls are the hard part. In testing, the walls are usually not what fails. Three things decide whether a shelter performs:

• The door assembly. ICC 500 requires the door — not just the shelter — to take the 15-lb missile at 100 mph without penetration and to still open afterward. A shelter you cannot exit after the storm is a serious hazard in its own right, since debris routinely piles on top of doors. This is why tested assemblies use heavy-gauge steel leaves, reinforced frames, and multi-point hardware.
• Anchorage. An above-ground unit that is not engineered and bolted to an adequate slab can be moved or overturned as a whole, regardless of how strong the box is. In-ground units have their own uplift consideration.
• Siting and drainage. An in-ground shelter placed in a low spot with no drainage plan solves the wind problem and creates a water problem. We cover this in detail in do underground storm shelters flood?

None of this is exotic or expensive to verify. It is paperwork the manufacturer either has or does not have, and the entire question resolves in one email. If you are still deciding on type and material, our guides on above ground vs. below ground and concrete vs. steel vs. fiberglass cover the tradeoffs, and the cost breakdown covers what a compliant unit actually runs.

The nine, in one table