How Accurate Are Medium Readings?
How Accurate Are Medium Readings
How accurate are medium readings is the question that sits underneath every other question about mediumship, and it is also the question most often answered badly. A medium may quote a percentage, a client may quote a feeling, a skeptic may quote a single failed test, and none of those numbers means very much on its own. Accuracy in a medium reading is not one thing. It is a bundle of different things: whether a name was right, whether a detail could be checked, whether the message helped, and whether the reading could have been produced without any contact with the dead at all. This page takes that bundle apart, piece by piece, so that you can judge a reading by standards that actually fit what a reading is.
The first point to settle is the simplest and the most often skipped. There is no verified overall accuracy figure for medium readings, and anyone who gives you one is inventing it. No study has measured all mediums, all readings and all sitters and produced a single number, and no honest practitioner claims otherwise. What exists instead is a collection of smaller facts: individual readings that can be checked, individual mediums whose records can be examined, and a long history of tests, some passed and some failed, that show how hard the question is to answer. This page uses those smaller facts, and it leaves out any number it cannot support.
The second point is about what accuracy means to the person in the chair. A reading can be accurate in the narrow sense, a name, a date, a habit all correct, and still leave the sitter cold, because the message did not touch the question that mattered. A reading can also be wrong in several details and still bring real comfort, because one true detail landed where it was needed. Both outcomes happen, and a fair judgment needs to hold both at once. Accuracy of fact and value of the experience are related, but they are not the same thing, and confusing them is how most arguments about mediumship go wrong.
What Counts as Evidence, and What Does Not
The strongest evidence in a reading is a specific detail that you can check, that is unlikely to be guessed, and that you did not give to the medium first. A full name that is uncommon in your family, a nickname used by only one person, a habit tied to a particular room in a particular house, a phrase the person used often, a date that matters for a reason only you know, these are the details that carry weight. The test is not whether the detail feels meaningful. The test is whether it is true, whether it is specific, and whether the medium could have known it from ordinary sources. A detail that passes all three tests is good evidence for that reading. A detail that fails any one of them is not.
General statements do not count as evidence, however warm they feel. A medium who tells you that the person loved you, that the person is at peace, that the person is proud of you, or that the person is watching over you has said something true for almost every grieving person, and something that cannot be checked in any case. These statements can bring comfort, and comfort has real value, but they should not be counted as proof that a connection happened. The same applies to statements about the future: a prediction that you will find peace, or that a difficult time will pass, is a general encouragement, not a testable claim. Count it as kindness, not as evidence.
There is a middle category that needs care. A medium may give a detail that is true but also easy to guess: a common first name, a cause of death that is common for the age, a love of music, a garden, a dog. These details feel personal because they belong to a real person, but they also belong to thousands of other real people. A single easy detail proves little. A run of them, ten or twelve in a row, each fitting the same person and none fitting anyone else in the family, starts to carry weight, because the chance of guessing a whole set is far lower than the chance of guessing one item. That is why notes matter: a run of details can only be judged from a written record, not from a feeling afterward.
The Houdini Test, and What It Taught
The most famous test of a medium in history is also the clearest lesson in how to judge accuracy. Harry Houdini, the escape artist, spent the last years of his life exposing mediums he believed were using tricks, and after his mother died in 1913 he sat with mediums himself, hoping for real contact and finding methods he recognized from his own stage work. Before his death on October 31, 1926, at the age of 52, Houdini and his wife Bess agreed on a secret code, a phrase only the two of them knew, so that if a message ever came through after his death, Bess would know it was really from him. The code was built from the words Rosabelle, answer, tell, pray, answer, look, tell, answer, answer, tell, a phrase tied to a song Bess sang in her stage days.
For ten Halloweens after that, Bess sat for a seance on the anniversary itself, October 31, waiting for the code to come through. In 1929, three years into the wait, the medium Arthur Ford claimed the code had come through to him. Bess believed him at first and signed a paper saying so. She later changed her mind, saying Houdini had sent nothing of the kind. The tenth and last sitting, broadcast from the roof of Hollywood's Knickerbocker Hotel on Halloween 1936, closed with Bess telling the audience the wait was over. No medium in ten years produced the code in a way that Bess accepted as genuine.
The lesson of the Houdini story is not that all mediums fail, and it is not that all mediums succeed. It is that a fair test needs a secret, a record, and a judge who is willing to say no. Houdini supplied the secret, the seances supplied the record, and Bess supplied the judgment, including the judgment that an early yes had been a mistake. Most readings have none of those three things: no secret agreed in advance, no written record made at the time, and no judge willing to mark an answer wrong. Without them, accuracy becomes a feeling, and a feeling is the easiest thing in the world to mistake for a fact. If you want to judge a reading fairly, borrow the shape of the Houdini test, even in a small way: agree on a checkable detail in advance if you can, write down what was said, and be willing to mark it wrong.
Why Readings Feel Accurate Even When Details Are Wrong
A reading can feel completely accurate while being wrong in most of its details, and the reasons for that are ordinary features of the human mind, not mysteries. The first is that grief makes the mind generous. A person who has just lost someone wants the reading to work, and that wanting fills in gaps, softens errors and turns a near miss into a hit. A medium who says a name close to the right name, or a place close to the right place, will often be helped by the sitter, who supplies the correction without noticing that they supplied it. The correction then feels like part of the reading, when it was really part of the sitter.
The second reason is the way memory works after a session. People remember the hits and forget the misses. A reading with twenty statements, five of them right and fifteen of them wrong, is often remembered as a reading with five right statements, because the wrong ones fade and the right ones are repeated to family afterward. This is not dishonesty. It is how memory works for everyone, in every kind of conversation. The only cure is a written record made during the session, before the forgetting starts. If you did not take notes, you cannot fairly judge the accuracy of a reading, because the reading you are judging is the one you remember, not the one that happened.
The third reason is the skill of the medium in reading the living person in the room. A good medium watches age, clothing, accent, wedding rings, hesitation, tears, and the small reactions that follow each statement, and adjusts the next statement to fit. This is often called cold reading, and it is a real skill, whether the medium uses it knowingly or not. The psychic vs medium distinction helps here: a psychic reads the living, a medium works with the dead. A statement that starts general and narrows as the sitter reacts can feel like a direct message from the dead, when it is really a careful reading of the living. That does not prove that no medium ever contacts the dead. It proves that a reading needs to be judged on details the medium could not have picked up from the room, and that a reading conducted with no information, no reactions and no feedback is a far stronger test than one conducted face to face with a grieving sitter helping along.
Evidential Mediumship and the Higher Standard
Within mediumship itself there is a higher standard, and it has a name: evidential mediumship. A medium working in this style aims to give evidence first and message second, and the evidence must be specific enough to identify the person: a name, a relationship, a cause of death, a shared memory, a habit, a place. Only when the sitter has accepted the evidence as correct does the medium move on to a message. The evidential mediumship page explains this approach in full, including its roots in the older Spiritualist movement, where the demand for evidence was a response to fraud.
The evidential standard is useful for a sitter because it turns a vague feeling into a checkable list. You can ask a medium before booking whether they work in this way, and you can ask during the session for the evidence to come before the message. A medium who welcomes that request is showing confidence in the method. A medium who refuses, or who says that evidence is not important, is telling you that the reading will be judged on feeling alone, and you should decide in advance whether that is what you want. Feeling alone is not worthless, but it is not evidence, and it should not be sold as evidence.
A Practical Way to Score a Reading Yourself
A simple scoring habit, used the day after a reading, gives a fairer verdict than any feeling in the room. Take your notes, or the recording if you have one, and go through each statement the medium made. Mark each one as right, wrong, or too general to judge. Count only the statements that were specific enough to be right or wrong. Ignore the general ones for the score, though you may still value them for comfort. Then look at the result. A reading where most specific statements were right, including one held soon after a death, is a strong reading. A reading where most specific statements were wrong is a weak reading, whatever the feeling was. A reading with very few specific statements at all is not a strong or a weak reading. It is an untested reading, and it should not be quoted as proof of anything.
This method also protects you from the two most common errors. The first error is counting a general statement as a hit, which inflates the score. The second error is forgetting the misses, which inflates it again. Writing the score down, with the notes beside it, stops both errors, because the paper does not forget and the paper does not flatter. Show the notes to a friend if you want a second view. A person who was not in the room will often see the general statements more clearly than you can, and will spot a detail you marked right that was really a near miss.
Finally, judge the medium by the whole pattern, not by one session. A medium who is right in specific detail across several readings, with different sitters, is worth returning to. A medium who is right once and vague ever after is not. A medium who is wrong in detail but honest about it, who says when nothing is coming through, is worth more than a medium who is never wrong, because a medium who is never wrong is not reporting what arrives. They are performing. The what to ask a medium page lists the questions to put to a medium before you book, and the how a medium reading works page explains the normal shape of a session, so you can tell the difference between a slow start and a stalled one.
For a simple next step, try a One-Card Tarot reading and reflect on which detail from your last reading you can actually check. First 3 minutes FREE — every time, for every reading, with all advisors.
Frequently Asked Questions
Is there a proven accuracy rate for medium readings?
No. There is no verified overall accuracy figure for medium readings, and any percentage quoted as a general rate is invented. Individual readings can be checked statement by statement, but no study has measured all mediums and all readings to produce a single number.
What is the strongest evidence in a medium reading?
A specific detail you can check, that is unlikely to be guessed, and that you did not give to the medium first, such as an uncommon name, a private nickname, a habit tied to a particular place, or a phrase only that person used.
Why did the Houdini seances matter for testing mediums?
Houdini and his wife Bess agreed on a secret code before his death on October 31, 1926, so Bess would know a genuine message. Bess held a seance each year for ten years, and the final one on October 31, 1936 ended with no accepted message, which shows the value of a secret, a record, and a judge willing to say no.
Can a reading feel accurate even if the details are wrong?
Yes. Grief makes the mind generous, memory keeps the hits and drops the misses, and a skilled medium reads reactions in the room. That is why notes or a recording made during the session are the only fair basis for judging accuracy.
How can I score a reading myself?
Write down every specific statement, mark each one right, wrong, or too general to judge, and count only the specific ones. A reading where most specific statements were right is strong, one where most were wrong is weak, and one with almost no specific statements is untested.
Last updated: November 11, 2026
