Loading field map
Defensive play in mahjong is the practice of adjusting one's own discards and strategy to reduce the chance that an opponent will win, especially when an opponent is close to completing a hand. It stands in contrast to offensive play, which focuses on building one's own hand toward a win as quickly as possible. The central question of defensive play is not "How do I win?" but "How do I avoid losing, and when should I stop trying to win in order to do so?"
Mahjong is a game of incomplete information. Each player sees only their own tiles, the discards of all players, and the melds (called pon, chi, and kan) that opponents have exposed. From this, a skilled player can infer a great deal about what tiles an opponent needs, but never with certainty. Defensive play is the art of managing that uncertainty.
The fundamental risk in mahjong is the deal-in: discarding a tile that completes an opponent's hand, causing them to win on your discard. In most scoring rules, the player who discards the winning tile pays the full value of the hand (or a large share of it), while the other non-winning players pay much less or nothing. This asymmetry makes defensive play strategically crucial: a single careless discard can cost more points than several rounds of careful building.
The key information available for defense is the discard pile. Every tile a player discards is a tile they did not need at that moment. By tracking which tiles each opponent has discarded, and in what order, a defender can build a picture of what that opponent is likely holding and seeking. For example, a player who has discarded all of their sou (bamboo) tiles is unlikely to be waiting on a sou tile to complete their hand. Conversely, a player who has kept many man (character) tiles and discarded other suits is likely building a hand concentrated in that suit.
Defensive play is not a single technique but a family of strategies that differ in how aggressively they sacrifice offensive potential and in how much certainty they require before acting. These approaches are not rival schools that displaced one another; rather, they form a spectrum, and skilled players move along it depending on the game state.
The most basic defensive approach is to discard only tiles that are known to be safe. A tile is safe against a given opponent if it cannot possibly complete their hand. The most reliable safe tiles are those that the opponent has already discarded themselves, because a player cannot have two copies of the same tile in a standard hand (there are four copies of each tile in the set, and a player's hand holds thirteen tiles plus one drawn). If an opponent discarded a 3-man earlier, they cannot now be waiting on 3-man to complete a pair or a sequence that includes it, because they would have needed that tile then.
A slightly weaker but still useful category is the genbutsu: a tile that was discarded by the opponent in question. This is the gold standard of safety. Beyond that, a player can use the furiten rule, which states that a player cannot win on a tile they have previously discarded (in most rule sets, this restriction is permanent for the hand unless they change their wait). Therefore, any tile in an opponent's discard pile is absolutely safe against that opponent.
The limitation of the safe tile method is that it is passive. A player who only discards tiles already thrown by opponents will often be unable to advance their own hand, and may be forced to break up their own promising tiles. Moreover, safe tiles run out. A hand lasts at most eighteen turns per player (in a standard game), and a player who has not been discarding useful tiles may find themselves with no safe discards late in the hand.
A more sophisticated approach involves tracking the flow of suits and honor tiles across all players' discards. This method does not aim for absolute safety but for probabilistic risk assessment. The defender asks: "Given what this opponent has discarded, what tiles are they likely to be waiting on, and which of my tiles are least likely to complete their hand?"
Key signals include:
The suji method is powerful because it allows a defender to identify tiles that are probably safe even when the opponent has not discarded them. However, it is not absolute. A player waiting on a pair (e.g., holding two 7-sou and needing a third) is not covered by suji logic. A player waiting on a kanchan (closed wait, e.g., holding 4-man and 6-man and needing 5-man) is also not covered. Skilled players deliberately exploit this by constructing waits that are "off-suji" to make their hands harder to defend against.
A refinement of suji is the kabe method, which identifies tiles that are not just relatively safe but nearly certain to be safe because all four copies are accounted for. If a player has discarded two 5-man, and you hold the other two 5-man in your hand, then no opponent can be waiting on 5-man (there are none left in the wall). More subtly, if a player has discarded 4-man and 6-man, they cannot be waiting on 5-man to complete a sequence, because they would have needed either the 4 or the 6 to make that sequence. This creates a "wall" that blocks certain waits.
The kabe method is a form of negative inference: it tells you what an opponent cannot be waiting on, rather than what they are waiting on. It is most useful in the late game, when many tiles have been discarded and the wall is dense. Its limitation is that it requires a large amount of information and does not cover all wait types.
A proactive defensive technique is sakigiri: discarding a tile that is likely to be dangerous before it becomes dangerous. This is based on the observation that a tile is safest when it is discarded early in the hand, before opponents have had a chance to build their hands around it. For example, if you draw a dora (a valuable tile that adds points to any hand) that you cannot use, discarding it immediately is safer than holding it and discarding it later, because by then an opponent may have built a wait that includes it.
The related concept of kiri (cutting) refers to the discipline of discarding tiles that are "dead" — tiles that cannot help your hand and are also dangerous to hold. Holding dangerous tiles is a common mistake: players keep a tile because it might become useful, only to be forced to discard it at the worst possible moment. Defensive play requires recognizing when a tile has no future in your hand and cutting it early.
The most extreme defensive stance is betaori, which means "folding" or "giving up on your hand entirely." When an opponent declares riichi (a ready hand, one tile away from winning) and you judge that your own hand is too weak to win quickly, you abandon all offensive plans. From that point, you discard only tiles that are safe against that opponent, even if this means breaking up your own hand completely.
Betaori is not a technique but a strategic decision. It is justified when the expected cost of dealing in (the probability of dealing in multiplied by the value of the opponent's hand) exceeds the expected gain of continuing to build your own hand. In tournament play, where point differences matter, betaori is often the correct choice even when your own hand is moderately strong, because a single deal-in can erase many rounds of careful point accumulation.
The limitation of betaori is that it is all-or-nothing. Once you fold, you cannot easily resume offense, and you may find yourself in a position where you have no safe discards left. Moreover, if multiple opponents are dangerous, folding against one may not protect you from another. Skilled players use betaori selectively, not reflexively.
These methods are not mutually exclusive. In practice, a player uses all of them in combination, shifting weight depending on the situation. Early in a hand, when no opponent has declared riichi, a player might use suit tracking and sakigiri to avoid creating dangerous tiles. After an opponent declares riichi, the player switches to suji and kabe analysis to identify the safest available discard. If the opponent's hand appears very valuable, the player may enter betaori and discard only genbutsu.
The relationship between these approaches is hierarchical. The safe tile method is the foundation: it provides absolute safety but limited coverage. Suji and kabe extend coverage by identifying probably-safe tiles. Suit tracking provides the context that makes suji and kabe more accurate. Sakigiri is a timing technique that makes all other methods more effective by reducing the number of dangerous tiles in your hand. Betaori is the strategic frame that decides when to deploy the others.
Defensive play as a systematic body of knowledge is a relatively modern development in mahjong's history. Early mahjong, as played in China in the late nineteenth and early twentieth centuries, was primarily an offensive game: players focused on building their own hands, and defense was largely a matter of intuition and experience. The game's spread to Japan in the early twentieth century, and the subsequent codification of Japanese mahjong (riichi mahjong) in the mid-twentieth century, created the conditions for more formal defensive theory.
The key catalyst was the riichi rule itself. In riichi mahjong, a player can declare riichi when their hand is one tile away from winning, placing a 1000-point bet and committing to discard every subsequent tile until they win. This declaration provides a massive amount of information: the opponent is ready, and their hand is now constrained. The riichi declaration made defense a central strategic problem, because it created a clear signal of danger and a clear moment to switch from offense to defense.
The development of suji theory is often attributed to Japanese professional players and writers in the mid-to-late twentieth century, though the exact origins are not well documented. What is clear is that by the 1970s and 1980s, Japanese mahjong literature had codified the core defensive concepts — genbutsu, suji, kabe, betaori — into a teachable framework. This framework spread to other mahjong variants, including Chinese official mahjong and Taiwanese mahjong, though the specific rules differ (for example, riichi does not exist in most Chinese variants, so the defensive calculus is different).
In the contemporary landscape, defensive play is a standard component of mahjong education. Online mahjong platforms, which allow for large-scale statistical analysis of games, have reinforced the importance of defense: data from high-level play consistently shows that players who fold appropriately against riichi declarations perform better over the long run than those who always push for their own win. However, the field is not static. Modern competitive play has seen a shift toward more nuanced risk management, where players do not automatically fold against every riichi but instead calculate the expected value of pushing versus folding based on hand value, turn number, and the score situation. This "expected value" approach does not replace the older defensive techniques but refines the decision of when to use them.
Defensive play is not a solved problem. The fundamental difficulty is that all defensive inferences are probabilistic, and skilled opponents actively work to defeat them. A player can construct a wait that is "off-suji" to make it harder to read. A player can discard tiles in a misleading order to suggest a different hand shape. The furiten rule, which normally protects a defender, can be exploited: a player in furiten cannot win, but they can still declare riichi to pressure opponents into folding, even though they themselves cannot win until they change their wait.
Another limitation is that defensive play is asymmetric across rule sets. In riichi mahjong, the furiten rule and the riichi declaration make defense more tractable. In Chinese official mahjong, where there is no riichi and the scoring system rewards a wider variety of hands, defensive information is sparser and the cost of dealing in is often lower, making aggressive play more viable. A player who learns defense in one variant cannot simply transfer all techniques to another.
Finally, defensive play has a psychological dimension that is difficult to formalize. Reading an opponent's tendencies — whether they are aggressive or conservative, whether they bluff with riichi, whether they hold dangerous tiles too long — is a skill that complements the mathematical methods. This aspect of defense remains closer to an art than a science, and it is one reason why experienced players can outperform purely statistical approaches in live play.