Remote Viewing and the CIA Stargate Files

What this piece gives you: the documented history of a real U.S. government program, what the two 1995 assessments actually said, and where the evidence stops. What it does not give you: an answer to whether remote viewing works. The files do not contain one.

Menlo Park, July 1974

Two physicists sat in that room: Harold Puthoff and Russell Targ, both out of laser research, both holding clearances from earlier work. Across from them, Pat Price, who claimed he could see places he had never been.

On the table were coordinates in degrees, minutes and seconds, plus a single hint: that this was a research and test facility. Only the client knew what was there, and the client was not in the room. Puthoff and Targ did not know either. That was the point of the arrangement — nobody talking to Price could leak anything, because nobody talking to Price knew anything.

Price described buildings, a site, and then something that held his attention: a multi-story crane running on rails, which he drew in detail. The sketch went to the client. The coordinates were Semipalatinsk in the Kazakh SSR, the Soviet nuclear test site.

That afternoon is the root of everything that followed: twenty-three years of research and operations, a succession of code names, roughly twenty million dollars, and a question nobody has answered. Not because the files are missing — they have been open since 1995, and the full CREST archive went online in 2017 — but because the same files support two different stories.

How a laser physicist ended up working for the CIA

Puthoff wrote the prehistory himself in 1996, in the Journal of Scientific Exploration, after the CIA released the first 270 pages of SRI reports in July 1995. It is a participant's account and should be read that way; it is also the most detailed access to the beginning, and it matches the files on the load-bearing dates.

In early 1972 a Puthoff research proposal on quantum biology was circulating. A copy reached the New York artist Ingo Swann, who wrote back suggesting he take up parapsychological experiments instead. Puthoff invited Swann to the institute for a week and arranged access to a heavily shielded magnetometer in Stanford's physics department. Swann, by Puthoff's account, perturbed the instrument and then drew its interior, though the design had never been published.

Weeks later two visitors turned up at Puthoff's office holding his report. Their credentials identified them as CIA. What drove them was not belief in the paranormal but a worry: Soviet security services were funding “psychotronic” research on a large scale. The Federation of American Scientists summarizes the assessment of the time as U.S. intelligence sources concluding, between 1969 and 1971, that Moscow was spending tens of millions of rubles a year on the field. Whether anything was behind it, nobody knew. But if there was, nobody wanted to find out last.

The visitors had Swann describe objects in sealed boxes while they watched. What followed was an eight-month pilot project worth $49,909, officially the Biofield Measurements Program. Targ joined. That is the start of what would later run as GRILL FLAME, CENTER LANE, SUN STREAK and finally STAR GATE.

The protocol — the most important part of the story

Most writing about remote viewing is writing about hits. That is understandable and it misses what Menlo Park was actually about. The real subject was not the hit. It was how you distinguish a hit from a coincidence, and how you rule out that the person drawing knew something they should not have.

The answer was a procedure with several partitions. In the Bay Area location experiments, a staff member drove to a randomly drawn target that only he knew. The viewer stayed in the lab with an experimenter who also did not know where the car had gone. Afterward the descriptions and sketches went to an independent judge, who had to match them to the possible targets without knowing which session belonged to which place. The series counted as a result only if that matching succeeded substantially more often than chance allows.

Role	Knows the target?	Receives
Client	Yes	The final transcript
Experimenter	No	Runs the session, tape recorder on
Viewer	No	Coordinates only
Independent judge	No	Transcript and sketches, matched blind against the target pool

How seriously the participants took this shows in one detail from Puthoff's account: the CIA monitors, alert to the possibility of trickery, served as viewers themselves to test the protocols. Three of them contributed seven of the fifty-five sessions in the series published in the Proceedings of the IEEE in 1976.

The pattern that emerges here runs through the whole history. The people who kept the program alive were not believers. They were skeptics who had seen something they could not account for.

Coordinates instead of a person: Scanate

The decisive step came from Swann. He proposed dropping the person at the target entirely and marking the site only by geographic coordinates. Puthoff calls it, in retrospect, a proposal that sounded even more absurd than ordinary remote viewing — anyone who has memorized a world map could guess. They called the procedure Scanate, scanning by coordinate.

In mid-1973 came the first test under outside control. A CIA threat-analysis office sent coordinates for a site neither experimenter nor viewer knew. Two viewers worked it, Swann and Price. According to the declassified final report, one drew a detailed map of buildings and grounds; the other supplied interior details including code words that were, in the report's phrasing, subsequently verified by the client's own sources. The site entered the files as the West Virginia Site. The intelligence chronicler James Bamford identified it as an NSA listening station at Sugar Grove.

What happened next is recorded without drama in Jessica Utts' 1995 assessment: one viewer named code words and personnel of the facility precisely enough to trigger a security investigation into how the information had gotten out.

Semipalatinsk: the crane with two truths

Which brings us back to that room in July 1974. The Semipalatinsk work ran in three phases — first blind, then with the clients participating while deliberately keeping themselves ignorant, finally attempting to generate material the CIA could not verify but would find operationally interesting. Over two weeks it produced several hours of tape and a notebook of drawings.

The evaluation, per the final report, found that several details of the essential technology matched other sources and that a number of specific large structural elements were correctly described. The results contained noise alongside signal but differed clearly from what control subjects produced in comparison experiments.

And here the puzzle starts — not with the crane, but with its assessment. Utts quotes a government analyst who reviewed the Semipalatinsk work. He concluded there was no way the viewer could have drawn the large gantry crane other than by actually having seen it through remote viewing, or by someone with knowledge of the facility telling him what to draw. The same analyst rated the session overall as unsuccessful — because the one hit on the crane was, in his words, far outweighed by the volume of incorrect material in the rest of it.

Both sentences are in the same document. The crane was a hit the reviewer could not explain, and the session was a failure. What you take from that depends on what you expect a session to deliver: one inexplicable hit, or a reliable picture. Utts' point is that you have to fix that expectation beforehand, not afterward — and that is exactly what the operational sessions lacked. There was no chance baseline to compare against, as there was in the laboratory series. And there was never, as she notes just as plainly, a way to rule out with certainty that the viewer had spoken to someone who knew the site, however unlikely that possibility seems.

Fort Meade: the experiment becomes a unit

Until 1975 the program was a small CIA-funded laboratory matter. Then the CIA withdrew — by Puthoff's account because of the congressional committees examining the intelligence agencies in the mid-seventies, for reasons that had nothing to do with remote viewing. Demand from other houses remained. In 1976 Edwin May, also a physicist, joined; a cadre of viewers was recruited.

In 1977 the Army began GONDOLA WISH, intended to assess what an adversary could do with such a capability. In 1978 that became GRILL FLAME: an operational unit in buildings 2560 and 2561 at Fort Meade, Maryland, staffed with soldiers and a few civilians thought to have some natural aptitude. In early 1979 the SRI research was folded in. In 1983 it was CENTER LANE. In April 1984 the columnist Jack Anderson made the unit's existence public — the first crack in the secrecy, ten years after Semipalatinsk.

Over the years more than forty people served in the program, including roughly twenty-three viewers. At its mid-eighties peak as many as seven were full-time. One of them, Joseph McMoneagle, states that he left in 1984 with the Legion of Merit for information on 150 targets unavailable from other sources. That is his account; the citation is not public.

When the Army exited in late 1985, the Defense Intelligence Agency took the unit over as SUN STREAK. In 1991 the research moved from SRI to the firm SAIC, and the whole thing acquired the name it is known by: STAR GATE.

The objection from New Zealand

Alongside Fort Meade another fight ran in the journals. In October 1974 Targ and Puthoff had published in Nature under the title Information transmission under conditions of sensory shielding. That a journal of that rank printed it was itself an event.

Four years later, in the same place, came the reply. David Marks and Richard Kammann of the University of Otago had reproduced the experiments. Their result, in one sentence: they had carried out duplicate experiments, and their results did not support the conclusions. Their central argument concerned the judges. The SRI transcripts, they showed, contained cues — references to earlier sessions, dates, incidental clauses — that let a judge reconstruct the order of sessions and therefore guess the matching, with nothing paranormal involved. Sensory leakage is the technical term.

That was not an accusation of fraud. It was a demonstration that one partition in the scheme was not sealed. And it is why the later protocols got stricter — which in turn is the reason Utts and Hyman agreed in 1995 on one point: the SAIC-era experiments were free of the weaknesses that had dogged the early SRI work. Both wrote that down. It is the last thing they agreed on.

1995: two reviewers, the same data, two answers

The 1995 defense appropriations process directed that the program be transferred to the CIA and evaluated retrospectively. The CIA commissioned the American Institutes for Research, and AIR brought in two reviewers: Jessica Utts, professor of statistics at the University of California, Davis, and Ray Hyman, professor of psychology at the University of Oregon and a long-standing critic of parapsychology. Both received the same material — essentially the SAIC experiments of 1992 to 1994 plus the SRI history. Both filed in September 1995.

Utts' first sentence after the summary is the one that has been quoted ever since: that by the standards applied in any other area of science, psychic functioning is well established. The statistical results, she wrote, lie far beyond what chance would predict. The effect is in the range social scientists call small to medium — small, but reproducible in cleanly designed series with enough trials. And she recommended running no further experiments designed merely to prove the effect exists: anyone who does not accept the existing data will not be helped by more. Study how it works instead.

Hyman's report is dated September 11, 1995. It opens by recording where he agrees with Utts: the SAIC experiments are free of the known error sources, and the effect sizes are too large and too consistent to be dismissed as statistical flukes. Then he separates three questions. Do these effects justify concluding that anomalous cognition exists? Is the possibility of methodological error fully excluded? Are the results consistent with those of other laboratories? His answer to all three: no.

His main argument is not that the data are wrong. It is that statistics alone establish no mechanism. Parapsychology, he argued, is unique in trying to demonstrate a phenomenon exclusively through departures from a chance baseline — by eliminating every other explanation, never through a positive model. Argue that way and all you have ever shown is that you have run out of alternatives.

Then comes the point the case turns on. Hyman wrote that, as he understood it, Edwin May — the program's director — had been the sole judge in all free-response experiments. May's reasoning: he knew the individual viewers' idiom. If one said “bridges,” May knew that viewer meant anything spanning water, and could score accordingly.

Hyman conceded the reasoning has sense to it. But it produces exactly the problem science has always avoided: a result that depends on one particular person. Evidence only one human being can generate is not evidence for anyone else. And on this single point, Hyman wrote explicitly, he and Utts agreed: the results must be obtainable with different judges.

So the whole dispute compresses into one question. Utts asks: are the numbers real? She answers yes. Hyman asks: can I rule out that they come from the procedure rather than the phenomenon? He answers no. Both are right. They are simply not answering the same question.

The report that ended it

AIR combined both assessments on September 29, 1995 and recommended terminating STAR GATE. The reasoning separated laboratory from operations cleanly: in the lab there are statistically significant results; a benefit to intelligence collection is not thereby justified. The CIA concluded there was not a single case in which remote viewing had delivered data that guided an intelligence operation.

What gets lost in that formulation: between 1986 and the first quarter of fiscal 1995 the unit received, per the AIR report, more than 200 taskings from operational military organizations wanting information unavailable from other sources. For two decades, cleared people with responsibility kept asking this unit. Whether they did so because they got results, or because asking cost nothing, the files do not say.

And then there is a former president's account. Jimmy Carter told students in Atlanta in September 1995, per Reuters, that during his term a plane had gone down in Zaire, satellites had not found the wreck, and the Director of Central Intelligence had turned — without his knowledge — to a woman said to have psychic abilities. She gave coordinates, the satellite cameras were pointed there, and the plane was there. That is a documented statement by a person, relayed by Puthoff from Reuters. No independent confirmation exists. The Federation of American Scientists lists the case among claimed successes, with the qualifier “allegedly.”

Claim, source, standing

Claim	Source	Standing
The program ran 1972–1995 under several code names and cost about $20 million.	Declassified files, FAS chronology, AIR report 1995	Verified
The laboratory results lie statistically well above chance.	Utts 1995 and Hyman 1995 — both	Verified
Early SRI transcripts contained cues that eased judges' matching.	Marks & Kammann, Nature 1978	Verified
Price described a gantry crane at Semipalatinsk that an analyst could not explain.	SRI final report, quoted in Puthoff 1996 and Utts 1995	Documented statement — the analyst rated the session overall a failure
A remote viewer located a plane downed in Zaire.	Jimmy Carter, September 1995, per Reuters	Documented statement by one person
Remote viewing works as a reliable capability.	—	Not established

What this means for you — and what it does not

You are probably not reading this as an intelligence historian, but because the question underneath it touches something: can a person perceive more than the senses deliver? The Stargate files give you no answer to that. They give you something better — a model of how honestly the question can be handled.

The Fort Meade viewers had a problem anyone recognizes who has tried to take their own intuition seriously. They called it analytical overlay: awareness catches a fragment — a shape, a texture, a sense of cold — and immediately starts building a story out of it. The fragment becomes a crane. Or a bridge. Or whatever you expected. The unit's entire methodology was an attempt to separate the fragment from the superstructure. And the 1995 argument was, at bottom, the same argument: what was signal, what was added?

That is not a case for a course promising you coordinate reading in three weeks. It is the opposite. It is a case for the exercise that precedes all perception: noticing what you perceive before you interpret it. Anyone who does not know their own interpretation machine cannot distinguish a signal from it — whatever the source. We have written that from the practice side in the clair senses explained and in opening the third eye.

What we do not promise

• That remote viewing is a reliable, learnable skill. Neither the files nor the 1995 assessments support that.
• That the program continues in secret. There are no documents for it, only assertions.
• That a course, a protocol or a practice produces accurate perception at a distance.
• That the unexplained hits prove anything. An unexplained hit is a hit nobody has explained yet.

Frequently asked questions

What was Project Stargate?

The last code name of a U.S. government program that from 1972 to 1995 researched and operationally used remote viewing. It began with CIA funding at the Stanford Research Institute, continued from 1978 as an Army unit at Fort Meade, passed to the Defense Intelligence Agency in 1986, and was terminated in 1995 after a review by the American Institutes for Research. Total cost: about $20 million.

Are the files public?

Yes. The CIA released the first 270 pages in July 1995, and the full CREST collection went online in 2017. The STARGATE material is searchable in the CIA's electronic reading room.

Did Utts and Hyman disagree about the data?

No. Both agreed the late laboratory results are too large and too consistent to be chance. They disagreed about what follows: Utts concluded the phenomenon is established; Hyman concluded that unexplained departures from chance are not evidence of a capability.

Why was the program shut down?

AIR recommended termination on September 29, 1995. The reasoning separated laboratory from operations: statistically significant lab results do not justify an intelligence benefit. The CIA found no case in which remote viewing had guided an operation.

What was the Semipalatinsk crane?

In July 1974 Pat Price described and drew a multi-story gantry crane at the Soviet nuclear test site. A government analyst wrote that he could not explain how it was drawn — and rated the same session unsuccessful, because the rest of it contained too much incorrect material.

Can I learn remote viewing from a course?

Nothing in the declassified record supports that. Hyman's 1995 report asked for replication with new viewers, new target pools and independent judges — a condition courses do not meet.

Sources and further reading

• Declassified CIA files, STARGATE collection, CIA electronic reading room (CREST), fully online since 2017.
• Jessica Utts, An Assessment of the Evidence for Psychic Functioning, 1995.
• Ray Hyman, Evaluation of Program on Anomalous Mental Phenomena, September 11, 1995.
• American Institutes for Research, final report, September 29, 1995; released November 28, 1995.
• R. Targ and H. Puthoff, “Information transmission under conditions of sensory shielding”, Nature 251 (1974).
• D. Marks and R. Kammann, “Information transmission in remote viewing experiments”, Nature 274 (1978), pp. 680–681.
• H. Puthoff, “CIA-Initiated Remote Viewing Program at Stanford Research Institute”, Journal of Scientific Exploration 10 (1996).
• Federation of American Scientists, program chronology and list of claimed successes.
• In the magazine: The clair senses explained · The morphic field, examined · What the near-death studies found · Opening the third eye
• Questions about an order or a product: 28@templeofdesire.com

This piece places a documented government program and its evaluations. It makes no claim of efficacy.