How to Choose Which Hand to Play in American Mahjong

The deal is done, the Charleston is over, and thirteen tiles are sitting on your rack. Somewhere on the NMJL card is the hand you should be playing — but the card holds dozens of hands, the clock is running, and nothing on your rack spells anything out neatly.

Choosing a hand is the central skill of American Mahjong. The good news: it follows a repeatable process, and the process matters far more than memorizing the card. This guide walks through it step by step.

Note: Example tile groupings below are illustrative fragments only — deliberately incomplete, never totaling 14 tiles, and not taken from any NMJL card.

Step 1: Find What Your Rack Is Already Good At

Don't start with the card. Start with your tiles. Sort the rack and look for clusters — concentrations the deal handed you for free:

• Repeated numbers — two or three copies of the same tile (44 or 888)
• A dominant suit — six or more tiles in Bams, Craks, or Dots
• Even or odd concentration — mostly 2s, 4s, 6s, 8s, or mostly 1s, 3s, 5s, 7s, 9s
• Consecutive numbers — a run like 4-5-6-7 across your rack
• Honor density — several Winds, Dragons, or Flowers
• Jokers — count them; they reshape everything below

Most racks have one or two genuine clusters and a lot of noise. The clusters are your raw material; the noise is your future discards.

Step 2: Map Clusters to Card Sections

Each cluster points to a neighborhood of the card, because the card's sections are organized around exactly these features:

• Even-number cluster → the 2468 section
• Odd-number cluster → 13579
• A run → Consecutive Run
• Repeated copies of one number → Any Like Numbers
• Threes, sixes, and nines → 369
• Winds and Dragons → Winds and Dragons
• Digits of the current year, or a Soap to serve as zero → Year hands
• Lots of natural pairs and few Jokers → Singles and Pairs

You're not picking a hand yet — you're picking one or two sections to read closely. This is what makes the card manageable mid-game: you only ever study the neighborhoods your rack points to. (If the sections themselves are still unfamiliar, How to Read Your NMJL Card covers the layout.)

Step 3: Shortlist Two or Three Hands

Within your target sections, find the specific hands your tiles fit best, and count honestly: for each candidate, how many of its 14 tiles do you already hold?

A useful benchmark right after the Charleston: holding 7 to 9 of a hand's tiles is a strong position; 5 to 6 is playable; below that, the hand is a long shot unless your Jokers make up the gap.

The best shortlists contain hands that share tiles — often neighbors in the same section. When two candidates overlap on eight tiles, every useful draw advances both, and you can defer the final decision. When your candidates share nothing, every draw forces you to favor one and weaken the other.

Step 4: Weigh the Factors That Separate the Candidates

When the tile counts are close, four considerations break the tie:

Pairs are the bottleneck. Jokers can never fill a pair, so every pair in a hand's pattern must be collected naturally — and with only four copies of each tile in the set, a needed pair you don't already hold half of is the slowest part of any hand. Given equal counts, favor the hand whose pairs you already own. A hand needing FF is far more attractive when two Flowers are already on your rack.

Jokers favor big groups. The opposite logic applies to pungs, kongs, and quints: Jokers complete them happily. A rack with three Jokers can chase hands full of large groups; a rack with none should lean toward hands it can finish naturally. Quints, remember, require Jokers — five of a kind doesn't exist in the set without one. (Full Joker rules: What Are Jokers in American Mahjong.)

Exposed beats concealed for beginners. Hands marked X let you call discards, which roughly adds the discard pile to your sources of tiles. Hands marked C must be built almost entirely from your own draws — slower and less forgiving. The concealed hands pay more for exactly this reason. Early in your playing career, the X hands are usually the right trade.

Value is the last tiebreaker, not the first. A 30-point hand you finish beats a 50-point hand you don't. Chase value only when the tile counts genuinely tie.

Step 5: Narrow as the Game Tells You To

Your shortlist isn't a decision — it's a holding pattern. The game itself makes the final call:

• A key draw or successful call that advances one candidate and not the others is the game voting. Listen to it.
• Watch the discards for your needed tiles. Only four copies of each tile exist. If three copies of a tile one candidate requires are already in the discard pile, that candidate just got much worse — and if all four are gone, it's dead. Cross it off.
• By mid-game, be down to one hand. Flexibility is an early-game virtue. Past the halfway point of the wall, splitting effort between two hands usually means finishing neither.

And give yourself permission to be wrong: sometimes the right move is abandoning the plan entirely when the draws head somewhere else. Knowing when to do that is its own skill — covered in When to Change Hands Mid-Game.

A Worked Habit, Not a Formula

The whole process compresses into a rhythm you'll run automatically within a few games: clusters → sections → shortlist → tiebreakers → narrow. The first few times it takes real effort and slows your turns down; soon it takes ten seconds at the deal and runs quietly in the background the rest of the game.

For practice without a table waiting on you, deal yourself hands in Play mode and run the process at your own pace — or put a confusing rack into the Hand Finder to check your shortlist against what the tiles actually support.