Tarot Spreads: Which One to Use, and What Waite Actually Wrote

This is a guide to choosing a spread rather than collecting them, plus a straight reading of what A. E. Waite actually wrote in 1910 — because on several points the modern teaching and the source disagree, and it is worth knowing which one you are following.

What a tarot spread actually does

A spread does one job: it decides what each card is being asked before you know which card it is. That ordering matters more than the shape. If you lay ten cards down and then decide what each position means, you have not read a spread — you have written a story around whatever turned up.

This is also why the position names are doing most of the work. "What crosses him" and "your obstacle" and "what is blocking you" are three different questions that will pull three different readings out of the same card. Before you adopt a spread, read its position names as questions and ask whether you actually want those questions answered.

Which tarot spread should you use?

Match the spread to the shape of the question, not to how important the question feels. A large spread on a small question produces padding; a small spread on a tangled question produces a shrug. This is the table we would actually use:

Note what is missing from that table: there is no row for "is he thinking about me". Spreads aimed at another person's inner life are the most-searched spreads there are and the least defensible ones — you are asking the cards about someone who did not consent to the reading, and you have no way to check the answer. We cover the reasons in how to prepare for a tarot reading.

The three-card tarot spread, and why the labels keep changing

Three cards is the workhorse. It is also the spread where the position names are least standardised, and that is a feature rather than a defect: the same three slots can carry at least a dozen different question sets. Choosing the labels is choosing the reading.

One rule keeps three-card spreads useful: fix the labels before you shuffle, say them out loud, and do not swap them afterwards because the cards would read better the other way round. That swap is the single most common way a reading stops telling you anything.

If you want to work with this yourself, this is what we use

Hermetic Tarot Deck — 78-Card Golden Dawn Deck With 72-Page Booklet

A full 78-card deck in the Golden Dawn line rather than the Rider–Waite–Smith one — worth knowing before you buy, because the court cards face differently and that changes where Waite’s positions 5 and 6 land.

$37.95

See the deck

The Celtic Cross tarot spread: what Waite actually wrote

The Celtic Cross is the most taught tarot spread in English, and it comes from one printed source: A. E. Waite's The Pictorial Key to the Tarot, 1910, where it is not called the Celtic Cross at all. Waite titles the section "An Ancient Celtic Method of Divination". The name we use is later.

The text is in the public domain and runs to about three pages. Read side by side with a modern diagram, four differences stand out — and three of them change how the spread is read.

Positions 3 and 4 are the substantive drift. Waite's pair is a time and reality distinction — what is hoped for and not yet real, against what is already in place. The modern conscious/subconscious pair is a psychological distinction that arrived decades later with the Jungian reading of tarot. Both are workable. They are not the same spread, and a card that reads clearly as "not yet actual" can read as nonsense under "conscious mind".

The significator: the card almost everyone dropped

Waite's spread uses eleven cards, not ten. Before any shuffling, the reader picks a court card to stand for the person or matter being asked about, lays it face up, and builds the whole spread on top of it. Card one is literally laid across it — that is what "this covers him" means.

Modern practice has quietly dropped it, and the reason is not mysterious. Waite's instructions for choosing one are a sorting rule by hair colour, eye colour and complexion: Wands for "very fair people, with yellow or auburn hair"; Pentacles for "very dark brown or black hair, dark eyes and sallow or swarthy complexions". It is a period artefact and it reads badly now, which is a fair reason to leave it out. What is worth knowing is that dropping it has a mechanical consequence, not just a cosmetic one.

If you want the significator without the complexion chart, Waite gives you the alternative himself: pick the card by subject rather than by person. His own worked example is a question about a lawsuit, where he takes Justice — trump eleven — as the significator because it corresponds to the matter. That version survives the century intact.

Why positions 5 and 6 have no fixed direction

Nearly every Celtic Cross diagram online fixes the past to the left and the future to the right. Waite does not. He places card five "on the side of the Significator from which he is looking" and card six "on the side that the Significator is facing". The court cards in the Rider–Waite–Smith deck face different directions, so the axis flips depending on which one you drew.

Waite anticipated the obvious objection and answered it in a note: if the significator is a trump or a small card that cannot be said to face either way, "the Diviner must decide before beginning the operation which side he will take it as facing". That is the tell. The direction was never a property of the layout — it was a property of the card, and the fallback was a decision made in advance.

Drop the significator and there is nothing left to decide the orientation, which is precisely why the modern spread had to nail past and future to fixed sides. That is a sensible fix. It is also a change, and if you have ever wondered why the past and future positions in the Celtic Cross feel arbitrary compared to the rest of the spread, this is why.

The re-draw rule nobody teaches

Waite closes the section with two instructions that almost never make it into modern write-ups, and both are useful:

• If card ten is too ambiguous to conclude anything, take card ten as the new significator, reshuffle, and lay the ten again. You are not asking twice. You are asking the same question at higher resolution.
• If card ten is a court card, that means the matter "falls ultimately into the hands of a person represented by that card" — so run the spread again with that court card as significator to see what they will do with it.

Compare this with the modern anxiety about re-drawing, where pulling again is treated as cheating. Waite built the second draw into the method. What he did not build in was re-drawing because you disliked the answer, and the difference between those two things is whether the first spread gave you a usable result. We have written separately about what to do when two readings contradict each other.

How many cards should a tarot spread have?

As few as answer the question. Card count buys detail, and detail costs coherence: every extra card is another thing to reconcile with the others, and past a certain point readers stop reading the spread and start negotiating with it.

The practical consequence: if you are new, the Celtic Cross is the worst possible starting point, and it is the one most beginner guides open with. Ten positions is ten unfamiliar questions plus seventy-eight unfamiliar cards. Start at one, move to three when one feels thin, and leave ten until the cards themselves have stopped being the hard part.

Can you make up your own tarot spread?

Yes, and it is a better exercise than collecting other people's. A spread is just a list of questions with an order, so writing one is writing the questions you actually have. Three constraints keep a homemade spread honest:

1. Every position is a question, not a topic. "Career" is a topic and will accept any card. "What am I avoiding at work" is a question and can be answered wrongly, which is what makes it worth asking.
2. No position may be redundant. If two positions would take the same card equally well, you have one position written twice.
3. Include at least one position you might not like the answer to. A spread built entirely from "what helps me" and "my strengths" cannot tell you anything you did not already want to hear.

Write the positions down before you shuffle, and keep them. A spread you invent and then reconstruct from memory next month is two different spreads.

When the card contradicts its position

This is the moment that decides whether someone keeps reading tarot. The Ten of Swords lands in "best possible outcome". The Sun lands in "the obstacle". The instinct is to declare the spread broken or to reshuffle.

The friction is usually the reading. A position is a question, and a card that sits oddly in it is answering that question with something you did not expect — which is the only circumstance in which a reading can tell you anything new. The Sun as obstacle is a legible answer: optimism, exposure, or someone else's good fortune is in the way. The Ten of Swords as best achievable outcome is legible too: the best available result is that a thing ends completely.

The exception worth naming: if most positions are fighting their cards, the question was probably wrong rather than the cards. Vague questions and questions about other people's minds both produce spreads that read like static.

Yes or no tarot spreads

These exist in volume and they are the weakest thing in the repertoire, for a structural reason. Tarot's whole apparatus — seventy-eight images, reversals, positional meaning — is built to describe a situation, and a yes/no spread throws all of it away to extract one bit of information. You are using a very fine instrument as a coin.

The common methods are worth knowing so you can judge them: counting upright versus reversed cards across three; assigning yes to one suit group and no to another; or reading a single card and calling anything "positive" a yes. The third is not a method, it is you deciding. The first two are at least consistent rules, but they rest on a reversal-as-negative assumption that most readers do not otherwise hold.

The reliable substitute is a three-card reframe: what happens if I do · what happens if I don't · what I am not seeing. It answers the decision the yes/no question was standing in for, and it survives being wrong.

Tarot spreads for love, career and decisions: what actually changes

Search autocomplete is dominated by topical spreads — love, career, relationships, the week ahead, moving house, a new job. It is worth saying plainly that in almost all of these, the spread is identical and only the framing sentence changes. "Situation, obstacle, advice" is the same three positions whether the situation is a relationship or a contract.

Two things genuinely do change by topic:

• Timeframe positions. A career spread benefits from an explicit horizon — "in this quarter" rather than "in the future" — because career questions have deadlines and love questions usually do not.
• Whose behaviour a position asks about. Relationship spreads slide towards reading the other person. Rewriting those positions as "what I am contributing" and "what I am reading into this" keeps the spread answerable.

So the honest advice is unpopular: you do not need a love tarot spread. You need the three-card spread with a love question and one position that asks about you.

Which spread fits the question you actually have?

Pick the sentence closest to what you would say out loud.

Voices from practice

We read through public beginner threads and reading forums while putting this together. We are not quoting anyone — these are the complaints that repeated, paraphrased:

• The Celtic Cross is the first spread most people are taught and the first one they abandon. The usual reason given is not that it is hard but that positions 3, 4 and 9 all seem to want the same information.
• Readers who stayed with it almost all describe the same fix: dropping to one or three cards for months, then coming back.
• The most common request in those threads is for a spread that will settle a question about another person's feelings — and the most common answer from experienced readers is a redirect to a spread about the asker instead.

What we don't promise

• No spread predicts events. Nothing here tells you what will happen, and position ten being called "what will come" is Waite's phrasing, not a claim we are making on his behalf.
• Waite's version is not more correct, only earlier. The modern Celtic Cross is a century of practice layered on a 1910 text. We have set them side by side so you can pick knowingly, not to award a winner.
• "Ancient Celtic" is a marketing name. There is no evidence of this layout in Celtic antiquity. It appears in print in 1910, in English, from a member of a Victorian occult society.
• Tarot is not advice for medical, legal or financial decisions, and a spread that seems to tell you to stop treatment or sign something is telling you what you brought to it.
• A deck does not make you a reader. The card stock is nice. The reading is the part you build.

Practically: how to work with this

1. Write the question down. One sentence, on paper, before the deck is touched. If you cannot write it in one sentence, the spread will not fix that.
2. Choose the spread from the question, not the mood. Use the table above. Most questions are three cards.
3. Write the position labels down too. These are the questions the cards are answering. Unwritten labels drift mid-reading.
4. Shuffle, cut, lay. Waite cuts three times. Anything consistent is fine; consistency is the point.
5. Read the positions in order, once, without conclusions. Name what each card is saying to its own question before you connect anything.
6. Then connect. Which cards agree, which fight, and what does the fight look like from outside.
7. Photograph it and leave it overnight if the question mattered.
8. Write the outcome down weeks later. This is the only step that ever improves anyone's reading, and it is the one nobody does.

Before you lay a spread with a deck that has just come out of storage, it is worth knowing how to cleanse tarot cards without damaging the cardstock.

Frequently asked questions about tarot spreads

Sources & further reading

• A. E. Waite, The Pictorial Key to the Tarot, London 1910 — Part III, § 7, “An Ancient Celtic Method of Divination”. Public domain; the position names, the significator rules and the re-draw instructions in this article are quoted from that section. Full text at the Internet Sacred Text Archive.
• A. E. Waite, The Pictorial Key to the Tarot, 1910 — Part III, § 6, “The Art of Tarot Divination”, for the shuffling and cutting procedure that precedes the spread.
• Pamela Colman Smith’s 1909 illustrations, which determine which way each court card faces and therefore which side Waite’s positions 5 and 6 fall on. We cover her authorship and the deck’s public-domain status in who drew the Rider–Waite tarot.
• Google Search autocomplete, US, September 2026, used to establish which spread questions are actually asked rather than which ones are commonly published.

Related reading: the 22 Major Arcana explained and the other 56 cards if the positions are clear but the cards are not; Rider–Waite vs Thoth vs Golden Dawn if you are still choosing a deck; and tarot vs oracle cards, since oracle decks mostly do not use positional spreads at all. Browse the tarot and divination shelf if you want a second deck to read against your first.