Polygraphs are unreliable because they don’t detect lies. They measure stress responses in the body, and stress can come from dozens of sources that have nothing to do with deception. Studies conducted outside the polygraph industry have found false positive rates as high as 50%, meaning half of truthful people can be flagged as liars. The National Research Council, after reviewing decades of research, concluded that the polygraph is “at best an imperfect instrument” and that practitioners’ claims of high accuracy “have rarely been reflected in empirical research.”
What a Polygraph Actually Measures
A polygraph records four things: heart rate, blood pressure, breathing rate, and sweat gland activity on the skin. These are all controlled by your autonomic nervous system, the part of your body that reacts automatically to stress, fear, surprise, or discomfort. The examiner asks a series of questions and watches for spikes in these signals, comparing your responses to “control” questions against responses to the questions that matter.
The core problem is that none of these signals are unique to lying. Your heart rate jumps when you’re anxious, angry, embarrassed, or simply afraid of being falsely accused. Your palms sweat when you’re nervous about the test itself. A truthful person who is scared of failing looks physiologically identical to a person who is actually lying. The polygraph has no way to tell the difference.
The False Positive Problem
A false positive happens when a truthful person is incorrectly labeled deceptive. Polygraph proponents often cite accuracy rates between 83% and 95%, but those numbers come from controlled lab settings where conditions are ideal. Independent studies, conducted outside the polygraph community, have found false positive rates of 50% or higher. That means in real-world conditions, a coin flip can be just as accurate at identifying honest people.
The math gets worse when polygraphs are used for security screening. The National Research Council calculated that when you’re screening a large population where actual threats are rare (say, 1 in 1,000 people), even an optimistic estimate of polygraph accuracy would produce hundreds of innocent people flagged as deceptive for every genuine threat detected. The spy or terrorist would be statistically indistinguishable from the flood of false positives. This is why mass screening with polygraphs is particularly problematic: the rarer the thing you’re looking for, the more the errors overwhelm the results.
People Can Beat the Test
Polygraphs are also vulnerable to deliberate manipulation. In a study where 80 guilty participants were trained in simple countermeasures, about half of them successfully fooled the test. The techniques were remarkably basic: biting the tongue, pressing toes against the floor, or silently counting backward by sevens during control questions. These small actions artificially inflate the body’s stress response during “safe” questions, making the guilty person’s chart look normal by comparison.
The countermeasures had their strongest effect on heart rate and blood pressure readings. More concerning, the researchers found that these techniques were difficult to detect either by watching the person or by analyzing the instrument data. If someone knows how the test works and prepares accordingly, the polygraph becomes significantly less useful, which means the people most motivated to deceive it (those with the most to hide) are also the most likely to beat it.
Examiner Judgment Adds More Uncertainty
Polygraph results aren’t just read off a machine like a thermometer. Examiners rely on a mix of subjective evaluation of the charts, partly objective numerical scoring methods, and sometimes computerized algorithms. The National Research Council noted that how well an examiner formulates the questions “inevitably affects the quality of information recorded.” The way a question is phrased, the tone of voice used, even the order of questions can all shape the physiological response.
Examiners also bring outside knowledge into their interpretation. They may know the person’s background, demographic information, or behavioral observations from the session. This introduces confirmation bias: if an examiner already suspects someone is lying, they may interpret ambiguous chart patterns as deceptive. Two examiners looking at the same data can reach different conclusions, which is a fundamental problem for any tool that claims to be scientific.
Medical Conditions Skew Results
People with conditions that affect their autonomic nervous system can produce misleading results in either direction. Someone whose body naturally produces blunted stress responses (due to nerve damage, certain medications, or neurological conditions) may show little variation between control and target questions. This can lead to inconclusive results or, worse, false negatives where a deceptive person passes the test because their body simply doesn’t react strongly enough to register.
On the other end, people with anxiety disorders, chronic pain, or cardiovascular conditions may show heightened responses to every question, making it nearly impossible to distinguish genuine deception from a body that’s in a constant state of physiological arousal. The polygraph treats all bodies as if they respond to stress in the same predictable way, and they don’t.
Why Courts Mostly Reject Polygraphs
Most U.S. courts do not admit polygraph results as evidence. For decades, courts excluded them under the standard set by Frye v. United States in 1923, which required that a scientific technique achieve “general acceptance” in its field before being used in court. Polygraphs never cleared that bar.
When the Supreme Court updated the standard for scientific evidence in 1993 with Daubert v. Merrell Dow Pharmaceuticals, it introduced new criteria: whether the technique has been tested, subjected to peer review, has a known error rate, and is generally accepted by scientists. Polygraphs struggle on nearly every count, particularly the high and variable error rate. A few circuit courts have since allowed polygraph evidence under narrow circumstances, such as when both parties agree to it in advance, but the overwhelming legal consensus treats polygraph results as too unreliable for the courtroom.
Where Polygraphs Are Still Used
Despite the scientific problems, polygraphs remain embedded in parts of the U.S. government. The FBI requires a polygraph examination as part of its background investigation for Top Secret security clearances. The CIA, NSA, and other intelligence agencies have similar requirements. In these contexts, the polygraph is used less as a definitive lie detector and more as an interrogation tool: the belief is that people who think the machine works will confess to things they otherwise wouldn’t.
In the private sector, the Employee Polygraph Protection Act prohibits most employers from using lie detector tests for hiring or during employment. You can’t be required to take one, and you can’t be fired or disciplined for refusing. There are exceptions for security firms (armored car companies, alarm services, guard companies), pharmaceutical manufacturers and distributors, and situations where an employee is reasonably suspected of involvement in a specific workplace theft or loss. But for most American workers, polygraph testing is illegal.
Why Accuracy Hasn’t Improved
One of the most striking findings from the National Research Council’s review is that polygraph accuracy claims have barely changed over the instrument’s entire history. Practitioners have always claimed extremely high accuracy, and empirical research has consistently found it to be lower. An analysis of the literature found that conclusions about polygraph accuracy have not changed substantially since the earliest studies, and the prospects for improvement “have not brightened over many decades.” The fundamental limitation isn’t the technology of the sensors. It’s that the underlying theory, the idea that lying produces a unique and detectable pattern of physiological arousal, has never been validated by science.
Newer approaches to deception detection, including brain imaging with EEG and fMRI, attempt to measure activity in the brain itself rather than peripheral stress responses. These technologies can distinguish between investigator-defined lies and truths with accuracy above 75% in controlled settings. But they share many of the same conceptual problems as the polygraph: lab conditions don’t replicate real-world stakes, individual variation is enormous, and the gap between “statistically better than chance” and “reliable enough to determine someone’s fate” remains wide.