Push or Fold in Mahjong: How to Price a Risky Discard

Push or fold is a price, not a mood

Play safe in mahjong when the cost of dealing in exceeds what your own hand is realistically worth. That comparison has three inputs – how close you are to winning, what your hand pays if it lands, and how dangerous the tile in front of you is – and every good fold decision comes from weighing those three, not from nerves.

Most players get this backwards. They push because they are already invested, and they fold because they feel uneasy. Reversing that habit is the fastest available improvement in the game.

The three inputs, in order

1. How close am I?

Count the tile swaps between your current hand and a finished one. Being one tile away is a different universe from being three tiles away, and the gap widens every turn as the wall shrinks.

Be honest about your wait as well as your distance. Waiting on a tile with only one live copy left is barely closer than not being ready at all.

2. What does it pay?

A hand worth the table minimum is not worth defending. If completing your hand gains you less than a typical deal-in would cost you, the maths has already decided – and it decides against pushing.

This is why knowing your ruleset matters. Wikipedia’s survey of scoring in mahjong is a useful reminder of how far apart the systems sit – Hong Kong, Taiwanese, Singaporean, Japanese and American tables all price the same shape differently, so a hand that is worth defending in one room is a giveaway in another. Our breakdown of basic scoring hands is worth reading precisely so you can price a hand mid-game rather than after it.

3. How dangerous is this specific tile?

Danger is not a general atmosphere; it attaches to individual tiles. A tile that three opponents have already discarded is close to free. A fresh middle tile late in the hand is close to a donation.

The decision matrix

Put the three inputs together and most situations resolve cleanly. Use this as a default and override it only for a specific reason.

Your handTile is clearly safeTile is unclearTile is obviously live
Ready, high value, wide waitPushPushPush, but only this turn
Ready, low valuePushFoldFold
One swap awayPushFold if two opponents look readyFold
Two or more swaps awayPushFoldFold
Any hand, and you are leading latePushFoldFold

Read the shape of that table rather than memorising the cells. Safe tiles are always playable, live tiles almost never are unless your hand is both finished and expensive, and everything in between depends on how many opponents are threatening you.

How to rank a tile’s danger without guessing

You do not need to read minds. You need a short list of cues, ranked from more to less reliable – and it is worth saying at the outset that none of them is a guarantee.

CategoryWhat it meansHow much to trust it
Already discarded by the threatThe dangerous player has thrown this exact tile themselvesStrongest under Japanese riichi rules; strong evidence rather than proof elsewhere
SujiThey discarded a 5, so under riichi rules a two-sided wait on 2 or on 8 is ruled outModerate – it covers two-sided waits only, never pairs or gap waits
Honours late in the handWinds and dragons others have already passed onModerate, and it drops sharply if nobody has discarded them

One important caveat: the discard-based cues are at their strongest in Japanese riichi, where the furiten rules restrict what a player is allowed to win on. Wikipedia’s article on Japanese mahjong sets those rules out in full, and they are worth reading there rather than second-hand from a strategy article.

Outside riichi that protection is weaker or absent, so at Hong Kong, Chinese and Taiwanese tables tracked discards are excellent evidence rather than proof. Suji reasoning leans on the same furiten logic, so it too is at its most reliable in riichi. Check your table’s house rules before relying on either.

Whichever cue you lean on, apply it to one player rather than to the table. The useful habit is to keep the threatening player’s discard row in view and read your own hand against that row specifically, instead of hunting for a tile that feels safe against everybody at once.

Three defensible reasons to keep pushing

Folding is undervalued at most tables, but a player who never pushes simply loses slowly instead of quickly. A push needs a reason you could say out loud, and only three of them survive scrutiny.

The wait is wide and the tiles are still out there

A finished hand waiting on two or three tile types with most copies unseen is a completely different asset from a finished hand waiting on the last of something. Count the live copies before you allow yourself to call the hand ready.

The payout can absorb being wrong

A hand worth several ordinary deal-ins can afford one mistake and still come out ahead over the night. That is arithmetic rather than bravado, and it is the only reason that holds up against a table full of visible threats.

The scoreboard has taken the decision away from you

Two situations change the currency entirely. Holding the dealership is worth pushing a marginal hand for, and so is needing points on the final hand that a minimum hand could never recover. Outside those two, the scoreboard argues for caution rather than against it.

A quiet table is permission, not a reason. If nobody has melded towards an obvious pattern, declared readiness or switched abruptly to conservative discards, a push costs less – but treat silence carefully, because an opponent who has stopped building looks exactly the same whether they have folded or finished.

Worked example: the last hand of the night

You are leading by a comfortable margin with one hand remaining. You are two tile swaps from being ready, your hand would score near the minimum, and the player to your left has melded two sets of bamboo and just discarded a wind they had been holding since the start.

You draw a 5 bamboo. It is the single most useful tile in the game for completing runs, and they are collecting bamboo.

  • Push: you keep a marginal chance at a hand worth almost nothing, and take a real chance of handing over the lead.
  • Fold: you discard something they have already passed on, you finish the hand with no points, and you win the night.

Fold, without hesitation. Your position is the reward here, not the hand. This is what people mean when they say strong players are protecting a result rather than chasing a win – the same long-run logic laid out in the argument that skill beats luck over time.

Three expensive mistakes

Folding without safe tiles in reserve

Deciding to fold on turn twelve with nothing safe in hand means folding in name only. Keep one or two known-safe tiles from mid-hand onward, even if it slightly slows your build.

Treating the whole table as one threat

Danger is player-specific. A tile that is suicide against the player on your right may be entirely safe against the other two, and folding against everyone at once is usually unnecessary.

Pushing because you have already invested

Tiles you have already discarded are gone whatever you do next. The only question that matters is what happens from this turn forward, which is the same trap covered in our list of mistakes new mahjong players make.

Turning your folds into feedback

Push-or-fold judgement improves through review, not volume alone. The cheapest habit available costs nothing: at the end of every hand you abandoned, look at what the winner was waiting on and at what you were holding, then ask whether the fold was actually necessary.

Physical tiles make that post-mortem possible in a way apps rarely do, because the discard rows stay laid out on the table after the hand ends and everyone can re-read them together. Three or four honest post-mortems teach more than a whole night of hands nobody examines.

The Mahjong Masters hires tile sets, tables and stools across Sydney if your group would rather practise on a real table than a phone screen. Bookings run as a day block rather than an hourly slot, a cash bond comes back when the set is returned intact, and every set is cleaned between hires – see what is available to hire before your group’s next session.

For the layer above this one – counting dead tiles, managing what your discards reveal – see our guide to advanced mahjong strategies.

Push-and-fold questions that come up at the table

When should I fold in mahjong?

Fold the moment finishing your own hand is worth less than the risk of completing somebody else’s. Distance and price do the deciding for you: a hand still several draws from ready, paying somewhere near the table minimum, is never worth a live tile against an opponent who is visibly committed.

What is the safest tile to discard?

A tile the dangerous player has already discarded themselves is the usual first choice, followed by honour tiles the rest of the table has already passed on. Neither is a guarantee outside Japanese riichi, where the furiten rules are set out explicitly.

Is playing defensively boring?

It is the opposite once you can see it. Folding well requires tracking every discard row and picking exit tiles several turns ahead, which is more demanding than pushing blindly.

Can I fold too much?

Yes. A night that ends with your score barely moved in either direction is a night of folds you did not need to make. Finished, expensive hands are meant to be pushed; only the marginal ones deserve the exit.

Does the same advice apply to every variant?

The framework does, but the safety signals differ. Discard-based safety is strongest under Japanese riichi rules because of furiten, while Hong Kong, Chinese and Taiwanese tables treat tracked discards as strong evidence rather than a guarantee.

How do I know an opponent is ready?

Watch for a change of behaviour: melds pointing at one obvious pattern, a sudden switch to safe discards, or a noticeably faster tempo. In riichi the declaration is explicit, but in most other rulesets you are reading the pattern.

Practice beats theory. Rent a full mahjong setup in Sydney and put these ideas in front of real opponents.

Share this

Previous Post

Mahjong for Team Building: Why It Works and How to Run It

Mahjong builds teams precisely because it is not a team game. Three session formats, a two-hour run sheet, kit per headcount, debrief questions and pitfalls.