The CIA surgically wired a live cat as a covert listening device, and its first spy mission ended when it was run over by a taxi

The story as it is usually told

The popular version is compact and unforgettable, and it is worth stating in its strongest form. After years of work and, by one widely repeated estimate, some twenty million dollars, the agency finally had a cat ready for the field. Technicians drove it to a park near the Soviet embassy, opened a van door, and released it toward two men on a bench.

In this telling, the animal made it only a few feet into the street before a passing taxi ran it down and killed it on the spot. The whole expensive experiment, the story goes, ended in a heartbeat under the wheels of a cab. The detail is delicious precisely because it is so humiliating: a fortune in Cold War ingenuity undone by ordinary city traffic before the cat could gather a single word.

And the source is not a random rumor. The account comes from Victor Marchetti, a former executive assistant to the CIA's Deputy Director, a genuine insider recounting what he described as an agency episode. Attached to a program the government has since confirmed was real, related by a former senior officer, the taxi ending arrives with real credentials.

A real project, a real insider, and a punchline too good to forget. The taxi story did not have to be invented from nothing; it grew on the back of something that actually happened.

That is the case at full strength: not that anyone has produced the cab's license plate, but that a credible former officer described a specific ending to a project we know existed, and that nothing about it is obviously impossible. Taken at face value, it is a believable capstone to a bizarre true story.

Where the anecdote thins out

Look closely at the ending, though, and the documentation supporting it nearly vanishes. The declassified “Views on Trained Cats” memo, the one primary record actually in the public domain, does not narrate a cat dying under a taxi. It discusses training limits and operational practicality and closes the project on those grounds. The most dramatic scene in the whole story appears in no released document at all.

What carries the taxi ending is essentially a single recollection, Marchetti's, offered years after the fact. A senior source is not nothing, but one person's memory of a covert episode, unaccompanied by any paper trail of the field test, is a thin foundation for a detail repeated as flat fact. The confidence of the retelling far outruns the evidence beneath it.

More decisively, another former CIA official contradicts it outright. In 2013, Robert Wallace, a former director of the agency's Office of Technical Service and co-author of the intelligence history Spycraft, said plainly that the cat was not run over. In his account, the surgical equipment was removed and the animal survived; the project was dropped not because of an accident but because a cat simply could not be reliably directed to go where handlers wanted. “The cat wanted to do what the cat wanted to do,” he put it.

That leaves two insiders telling incompatible stories, with no released document to arbitrate between them. When the sole support for a widely stated fact is one recollection, and a comparably placed official flatly denies it, the honest label for the taxi ending is not true and not false but unproven.

The twenty-million-dollar figure

The same thinness affects the number that makes the taxi story sting. The oft-cited twenty million dollars is, like the cab itself, attributed to Marchetti, and it appears in the released files as no confirmed line item. Because the surviving documents are heavily redacted, the program's true cost is not publicly verifiable.

This matters because the power of the legend depends on the pairing: an enormous sum destroyed in an instant of bad luck. Strip out the unverified total, or the unverified cab, and the anecdote loses its punch. Yet both halves of that memorable pairing come from the same lightly sourced account, which is exactly the circumstance in which a story should be held at arm's length.

A real program does not make every detail attached to it real. The surgery is documented; the price tag and the taxi are recollections, and recollections are not receipts.

None of this proves the figure is wrong or the cat unharmed. It means the confident, precise version, twenty million dollars, one cab, one dead cat, on the first mission, is a story built on a narrow base and repeated as though it were a settled account. The documented core deserves belief; the embellishments deserve a question mark.

Why the taxi ending endures

Acoustic Kitty is one of the most durable pieces of Cold War trivia going, and it survives for reasons that have little to do with whether a taxi was actually involved.

It endures because the premise is true. Most tall tales collapse the moment you check them; this one gets stronger, because the wired cat is real and the government released the proof. Once the outlandish setup survives scrutiny, people extend that trust to the whole package, taxi and all, without noticing that the ending was never on the same footing as the surgery.

It endures because it is a perfect little story. Expensive, secret, ambitious, and then flattened by ordinary traffic: the irony is so clean that the anecdote almost tells itself. Stories that good are copied faster than they are checked, and each retelling sands off the caveats until only the punchline remains.

And it endures because it confirms a suspicion people already hold, that vast, secretive agencies burn money on absurd schemes. The image of a doomed spy cat is a tidy parable of institutional folly, and a parable is not eager to be complicated by a second former officer insisting the cat lived. The satisfying version wins because it feels like a truth about power, even where the factual record is genuinely split.