Cheyenne Mountain Closed Its Doors For Real Four Times In Sixty Years
The doors themselves take under half a minute. The posture behind them takes twenty-two minutes on paper, and the internal exercise record shows that standard has not been met since 2014. The same two dampers are blamed every time.
COLORADO SPRINGS, United States — The two blast doors at the entrance to Cheyenne Mountain weigh 25 tonnes each and swing shut on rollers in 22 to 30 seconds. A crew that has done it before can beat that.
Since the complex was declared operational in 1966, they have been closed for a reason that was not a drill or a scheduled test on four occasions. We can date all four. Two of them are seven months apart.
The question we asked
Whether the mountain still matters is not a question with an evidentiary shape, so we did not ask it. We asked how many times the doors have actually been closed, which a chronology can answer.
Sixty years of unit chronologies, read against the quarterly exercise schedules, sort every closure into two piles: the ones printed on a schedule in advance, and the ones that were not. Forty-one of the sixty annual chronologies were released to us. Nineteen were withheld or could not be located, and that hole is the first thing a reader should know about the number four.
The four
9 November 1979. A training tape ran on a live system and produced a picture of an attack in progress. Chief Master Sgt. Ambrose Tillotson-Reyes, then a young controller, has a pocket diary that gives 08:53 for the doors going shut and 09:19 for them coming back. About six minutes of the button-up sequence had been completed.
3 June 1980. A failing circuit in a communications multiplexer produced a similar picture, twice in three days. The after-action note runs to two pages and is mostly about a telephone.
11 September 2001. Closed at 09:37, and on this occasion not reopened. The complex operated buttoned up for thirty-two days.
26 June 2012. Smoke from a fire on the far side of the ridge reached the air intakes. The doors were closed because of the intakes. Nobody was attacking anything.
Twenty-two minutes
The doors are the quick part, and the quick part is not the point.
What matters is the full button-up: dampers closed, filters in line, the plant switched over to internal air, the whole complex sealed on its own atmosphere. Twenty-two minutes, on paper.
The internal exercise record shows that time has not been met in any exercise since 2014. Forty-seven scored button-ups since then: best 24:10, in 2016; median a little over twenty-nine minutes; worst 41:22, in 2021.
Every after-action note in the run attributes the delay to the same thing, in nearly the same words — a pair of isolation dampers on the number-three air-handling line, named by asset number and never described.
What hardening is now
The register that would say what those dampers serve is withheld entire. The works programme that would fix them is not.
That programme is the actual state of the mountain in 2026: a queue. A 2019 survey of the 1,319 shock-isolation springs on which the fifteen internal buildings stand found 63 outside tolerance — a figure the survey gives and a tolerance it never defines. Remediation entered the five-year plan at position eleven. It is now at position nine, behind a fire-alarm replacement and a chilled-water valve renewal, both of which have slipped once themselves.
Nothing there is sinister. It is what an estate looks like when it is managed by a contract rather than defended by a garrison.
Where the work went
Day-to-day function left the mountain in 2006 for an ordinary building with ordinary walls seven miles away, and it did not leave because anybody decided granite was obsolete.
It left because the thing most likely to take the mission down now arrives on a fibre, and a fibre is not slowed by a door. Twenty-five tonnes of steel is an excellent answer to a blast wave and no answer at all to a routing table.
The sound
We found three men who have closed those doors for real, interviewed them separately in different months, and told none of them about the others.
All three, unprompted, described the same thing. Not the tonnage, not the granite, not the sirens. The sound the seal makes at the end, when the pressure comes up and the noise stops being mechanical.
One of them compared it to a refrigerator.
What we could not close
Brigadier General Rosalind Quennell-Adaeze read this piece before publication and sent back a fifth closure: October 1999, forty minutes, a telephoned threat from a man in Pueblo with a grievance about a child support order. We cannot find it in any record released to us. We have not added it to the count and we have not left it out of the paper.
Her larger objection is at the foot of this page and this desk has no answer to it. The twenty-two-minute standard, she says, describes a war the building is no longer expected to be the answer to — and the interesting fact is not that the mountain has run four to seven minutes late for twelve years.
It is that somebody has been out there with a stopwatch, every quarter, writing it down.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We did not ask whether the mountain still matters, because that question has no evidentiary shape. We asked a smaller one with an answer: how many times the doors have been closed for a reason other than a drill. That meant reading sixty years of unit chronologies against the quarterly exercise schedules so that every closure fell into one of two piles, and then finding men who had been standing there.
Who we spoke to
- Chief Master Sgt. (ret.) Ambrose Tillotson-Reyes, Command post controller, Cheyenne Mountain, 1976–1994. Interviewed at home in Fountain, Colorado, twice, and by telephone a third time May and June 2026 On duty on 9 November 1979. Kept a pocket diary that supplies the two clock times we print for that morning.
- Cheyenne Mountain historical office, Unit chronologies and quarterly exercise schedules, 1966–2026. Requested in writing over six months; 41 of 60 annual chronologies released, the rest withheld or not located October 2025 – April 2026 The chronologies are the only means we had of establishing that a closure was not on a schedule printed in advance.
- Base operating support contract, section J attachments, Facilities-maintenance contract and five-year works programme. Obtained in redacted form; the asset register is withheld entire, the works programme is not March 2026 Gives the queue position of every renewal, including the two items above the springs.
- Structural condition assessment of the spring assemblies, 2019 survey, 74 pages, 9 withheld. Released alongside the contract March 2026 Gives the figure of 63 springs outside tolerance and does not anywhere define the tolerance.
- Brig. Gen. (ret.) Rosalind Quennell-Adaeze, Deputy director of operations, Cheyenne Mountain, 1997–2003. Sent the full draft, the exercise table and the works programme; replied in writing after eleven days July 2026 Says our count is an artefact of filing and our headline metric is obsolete. Supplies a closure we did not have. Printed in full.
What we could not confirm
- Whether four is the count, or only the count that generated paper. We counted closures that appear in a chronology or an after-action note. A closure ordered and reversed inside two minutes need not produce either. Two of the three men we interviewed believe there was at least one more in the middle 1980s; neither can date it, and Brig. Gen. Quennell-Adaeze supplies a fifth, in 1999, that we could not find in any released record and have not counted.
- What the two dampers are. Every after-action note since 2014 names the same pair by asset number and never by description. The register that would say what they serve is withheld entire, and we cannot tell from the copy we hold whether that is a security decision or the way the pages were scanned.
- Where the twenty-two minutes comes from. The figure appears on the exercise scoring sheets, on a 1998 briefing slide, and in the mouth of every person we asked. Nobody could produce the order that set it, and one retired officer told us he had always assumed it began as somebody's stopwatch.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
The Meridian TelegraphTHE DOOMSDAY DOOR HAS FAILED EVERY TEST SINCE 2014
Took the run of missed times as evidence of decline, and the decline as evidence of somebody's intention.
Had that we did not
The twenty-two-minute standard, correctly quoted, and the twelve-year run of misses, which is real and which we confirm.
Left out
That the misses occur at a step the doors are not part of, and that the fix on the works programme is a linkage and two dampers.
-
Signal & LedgerA 1,319-Spring Estate: What Deferred Renewal Costs Inside A Mountain
Read the complex as a facilities portfolio and drew a lesson about condition surveys for readers who manage estates.
Had that we did not
The ranking logic of the five-year plan, explained more clearly than we explain it.
Left out
That two of the line items sitting above the springs have themselves been deferred, which their own table shows and their text does not say.
-
They Buriedthis newspaperCheyenne Mountain Closed Its Doors For Real Four Times In Sixty Years
Asked how many times the doors have actually been closed, and then asked what closing them is now for.
Had that we did not
All four occasions with their dates and times, every scored exercise since 2014, and the works programme in the order it is queued.
Left out
We asked three men what it was like and printed the sentence about the refrigerator because it was charming. Two of them also told us what they had thought was happening outside while the seal came down, and we cut it for length. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Brig. Gen. (ret.) Rosalind Quennell-Adaeze, deputy director of operations, Cheyenne Mountain, 1997–2003 on 6 July 2026, in writing, with the full draft, the exercise table and the works programme attached, and no deadline imposed. Replied 17 July 2026. Printed entire and unedited, including the parts about this newspaper.
Your arithmetic is sound and your question is the wrong size, and I would like to take longer over the difference than you will want to give me.
Four is not a number about the mountain. It is a number about filing. On an evening in October 1999 those doors were closed on my watch because of a telephoned threat that turned out to be a man in Pueblo with a grievance about a child support order. It lasted a little over forty minutes. Nobody was hurt, nothing was found, and I am reasonably confident it appears in no document you are able to request, because the reason expired inside the shift and the shift wrote up what had actually happened, which was nothing. So what you have published is a claim about archives wearing the clothes of a claim about history. I do not think you did that dishonestly. I think you did it because four is a better sentence than 'four that we can see'.
On your twenty-two minutes I am blunter. That figure was built around a design case in which the warning was counted in minutes and the threat arrived on a ballistic arc. Somewhere between 1991 and 1994 that stopped being the design case, and nobody wrote that down either, because institutions record the beginnings of things and almost never the ends. You have found a facility failing to meet a standard that describes a war it is no longer expected to be the answer to. That is not a decline. It is a metric outliving its reason, which happens in every organisation on earth and is only interesting here because the building is photogenic.
You give four paragraphs to two dampers. They are dampers. Eleven inches of plate and a linkage that has been rebuilt twice, made by one firm in Ohio with a fifty-one-week lead time, and the works programme says so on the page you reproduced. I am not defending the queue — I have views about the queue, and I said them in writing in 2002 and was thanked for them — but a fire-alarm renewal sitting above a spring assembly is not a scandal. It is a fire alarm. I would have put it above the spring myself. Fire is the thing that has actually killed people in that complex.
I have read what your three men told you and it is true and well observed and you asked them the wrong thing. You asked what it was like. Nobody has ever asked me what it was like, and if they had I would have told you about the sound as well, because the sound is the only part of it that is a story. What I would rather have been asked is what I was deciding while it came down. The door is not the decision. The door is what you do with your hands while you decide.
Where you are right, you are righter than you appear to know. The mission's exposure is on a fibre and a door does not slow a fibre by one nanosecond. If you want the frightening version of your own story, it is not that the doors are slow. It is that the doors are irrelevant and we kept measuring them anyway, for twelve years, four to seven minutes late, and filed it.
Published unedited under our right-of-reply guarantee.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.