A Small Canasta Strategy Guide

Back to the main page

Sending me mail works now!!!

(If you're about to ask me about the rules of canasta, please read this first.)

Rules of Canasta (John McLeod's site)



This document is based on my personal experience playing Canasta with experienced but not very seriously playing people. (My playing skills are based - in addition to experience - on a chapter about Canasta in a book called "MMM Korttipelikirja". That book has also influenced this document.) The goal of the document is twofold: First I hope that it helps novice players, but also I try to expose my views about the game so that experienced players can compare the way I see the game to their own way. If you do this, please mail me the results of your comparison. Also other comments - especially criticism - are welcome. English is not my mother tongue; corrections to mistakes with the language are welcome.

I try to explain what kind of things you must take into account when you plan your actions rather than give concrete advice. I think it's typical for Canasta that almost every decision has both good and bad sides, and you must carefully consider which ones are better in the particular situation that you're playing. I also try to explain the reasons why you should or should not do certain things. My advice is not perfect and you're strongly encouraged to deviate from it when needed.

And finally, this document covers only the 4-player game in 2-player teams with the classic 1950's rules.


The Nature of the Game

In each hand, usually one partnership is dominant and the other one is submissive. The dominant partnership has many cards, and thus many chances to take the discard pile, and many chances to make canastas. The submissive partnership has few cards, and few chances to do anything.

The partnership that takes one big or a couple of small discard piles becomes dominant. Thus, in the beginning of a hand, you should aim at taking many cards from the discard pile. (Since Canasta is a team game, the aim is getting as many cards as possible from the pile for your team. Whether you or your partner gets the cards is quite irrelevant. However, in the ideal position both you and your partner have many cards in your hands. This is discussed in depth later.) After deciding the dominance, the submissive team tries to go out and the dominant team tries to make the game last as long as possible.

This setting occurs almost certainly when the score required for opening meld is high (that is, at least 90). When the opening meld is 50 for both teams, there may appear also equal hands. In equal hands a continuous battle over the discard pile is going on. You must continuously consider which team wins more by continuing the game, and quickly go out if you think your opponents have better chances to complete canastas. I think there are no special "equal hand" strategies, but you must apply all strategies when they are suitable.

The Opening Meld

In the beginning, it is important that your partnership becomes the dominant one. This goal is achieved by taking one big or several small discard piles. Thus, the main reason for making the opening meld early is to improve your partnership's chances to get the discard pile. In some ways the opening meld makes it easier to get the pile, in other ways it makes it more difficult.

The ways making the opening meld makes it easier to get the pile are: (1) After the opening meld you and your partner can get the pile with one natural card and a wild card. (2) Your partner doesn't have to make the opening meld. The ways making the opening meld makes it more difficult to get the pile are: (3) The opponents see some of your cards. (4) You usually lose wild cards. (5) If the pile is frozen, your opponents will have safe discards.

When the score needed for the opening meld is low (that is: 50), the point 1 is the most important reason for making the opening meld. It's very probable that your partner has a decent opening meld in his hand, and he does not need your help for that. If you lose all your wild cards by making the opening meld, it's often not worthwhile. Because you don't then have any wild cards, half the advantages of the opening meld are lost. You'd better wait a couple of rounds and see if your partner can make the opening. Do not waste too many cards on the opening meld of 50 points. The opening meld of 50 points is easy, and after a couple of rounds either you or your partner will certainly have a nice 4-card opening.

The previous paragraph, of course, applies only to an unfrozen pile. If the pile is frozen and the score required for opening meld is low, you will probably gain nothing by making the opening meld.

When the score for the opening meld is high, it may happen that your partner cannot make the opening meld and he consequently has no chances to get the pile, unless you make the opening meld. In such cases, you should open if possible. In your first turn, make only an opening meld with a small number of cards, e.g. 90-point meld Joker-A-A. Otherwise you should wait and see if your partner can make an opening meld with a small number of cards. In your second turn, if your partner has not opened, you can make an opening meld with a little bigger number of cards, in your third turn with a still bigger number of cards etc...

To minimize the effect of points (3) and (5) you should make the opening meld with as little number of different combinations as possible, and with as little number of cards as possible. However, melding natural cards is often better than melding wild cards even if you have to use one or two cards more (especially if you meld longer - not more - combinations), because wild cards are much more valuable when they are in your hand than when they are on the table.

Your goal is to get the discard pile. If you can take a fairly big (e.g. 8 cards) discard pile by making the opening meld, you should do so. If the battle over dominance is lost (opponents have taken one big or a couple of small piles), you should open when it suits your submissive strategy.

Fighting over the Discard Pile

Small piles may not be worth taking. Taking a pile consisting of one or two cards makes your hand weaker because (1) you may have to meld cards in order to take a pile (2) you miss a chance to take a possibly good card from the draw deck (3) opponents know - and usually also remember later - which cards are in the discard pile, whereas they don't know which cards you get from the draw deck.

However, larger piles are almost always worth taking. The more cards you have in your hand, the stronger you are. Even though you may not need the cards in the pile, your opponents probably would need them, and taking them out of their reach is good strategy. The only case where you might not take a big pile is when you're planning to go out. Even then, the cards in the pile could possibly alter the situation so that continuing the game with those cards is better than going out with your current hand.

After this rather lengthy introduction, we finally turn to the subject of this chapter: How to take the discard pile.

There are two completely different cases of fighting over the discard pile: The frozen pile and the non-frozen pile. The principles of the fight are the same in both cases, but the way those principles are applied is different. The fight where one or both teams has not made the opening meld differs from the case where both teams have made the opening meld.

When you try to prevent the opponents from taking the pile, the main thing is to decide which cards are safe to discard: Black threes are always safe, and thus they are extremely valuable cards (this fact is very often neglected by novice players). Wild cards are also safe, but they freeze the pile. Otherwise, such cards that are improbable to exist in your left-hand opponent's hand are relatively safe. If the pile is frozen, the cards that fit your opponents' melds are relatively safe - the longer the meld the safer the card. In all cases, such cards that your team has many of (either melded or in your hand) are relatively safe, and such cards that your left-hand opponent has discarded are relatively safe (you mostly discard single cards.) Also such cards that you've discarded earlier and your left-hand-opponent has neglected are relatively safe. If your opponents have not opened, small cards are safer than large cards. (You try to get the opening meld by collecting large cards and discarding small ones.) That neglected cards are safe does not apply to a situation where your opponents are required a big opening meld that they have not yet made. It's possible that your left-hand opponent has not taken the pile because he could not open then, not because of the lack of the pair.

When the pile is small, you should discard unsafe cards and save the safe ones for big piles. This is because not much harm is done if your opponents get a small pile, whereas it's catastrophical if they get a big pile. Also, if your left-hand opponent does not take the pile with the unsafe cards you offered, you can consider those cards relatively safe later.

When you're defending the pile, the most important thing is to remember the cards you have discarded and the cards that your left-hand opponent has discarded.

The best way to get a discard pile is to have a pair of cards (if the pile is unfrozen, only single card with a wild card is enough) that your right-hand opponent assumes safe. You can - and you should - arrange yourself such pairs with cunning play: Do not meld everything you can. For instance, from five cards of same rank meld only three and leave a pair in your hand. If you have three cards of same rank, discard one and leave a pair in your hand. (This has also an additional advantage: If your left-hand opponent does not take the card you offered, you have two additional safe cards yourself). Do not take a small pile even though you could etc...

Remember also that the more pairs you have, the easier it is to get the pile and the more different cards you have (plus a wild card) the easier it is to get an unfrozen pile. However, it's not reasonable to discard from pairs in order to get more different cards: If your team loses the pile or the pile gets frozen, a hand that consists of single cards becomes totally worthless.

The discard pile is an essential part of the game, but it is not the entire game. If you eventually lose the battle over the pile, you should be able to go out as fast as possible. Many of the tricks advised above cause your hand to become such that it works well in the battle over the pile, but it does not work in meld-and-go-out play at all. Consider for example discarding from a combination of three or four cards of same rank. This will destroy a combination that you may need for making a canasta. You can not meld mere pairs.

So, you must find a balance between fighting over the pile and keeping your hand vital. The bigger the pile, the more you must fight over it, even though it makes your hand crippled for the normal game. Sometimes a frozen pile can become so big that the only goal for everyone is to get it, or at least prevent the opponents from getting it; skilled players can often defend it until the draw deck ends and neither side gets the pile.

Rememer also that the more cards you have in your hand, the stronger you are when you fight over the pile. Thus, stop melding when the battle over the pile gets fierce.

Submissive play

When your opponents have most of the cards, their chances of getting the discard pile again become bigger and their position probably becomes better and better. In such cases, your main goal is completing one canasta and then going out as fast as you can, before the opponents complete too many canastas. You'll probably lose the hand, but your goal now is minimizing the loss.

The other strategy you may use is freezing the pile and thus trying to break the submissive - dominant setting. For this strategy there is another chapter later.

The major decision you have to make is which cards you meld and which ones you keep concealed. The advantage to melding is that your partner can then use the cards: It's easier to build a canasta together with your partner than alone in your hand. If you already have a canasta, your melds can help the partner to get rid of single cards and go out.

Often your partner could go out if you had a canasta, but he can't both make the canasta and go out. Thus, often you should make a canasta in order to make it possible for your partner to go out. Usually both making canasta and melding pairs in order to get rid of cards in hand require wild cards. It's fastest to go out if one of you uses his wild cards to make canastas and the other uses his wild cards for melding pairs.

On the other hand, your opponents probably have many meldable combinations in their hands, and maybe they have some 5-6 card pure combinations melded which they try to turn into pure canastas. They try to guess when you will go out and meld their hands empty before that and make mixed canastas of those pure 5-6 card combinations. If you always meld everything you can, it'll become very easy for them to estimate how much time they have before you'll go out. If possible, you should try to go out surprisingly, and this requires hidden resources in your hand.

If you have to go submissive before you make the opening meld, then what is said in the previous two paragraphs about the melds in general applies also to the opening meld. Just remember that your partner may not be able to make the opening meld and you may have to make it in order to help him. When your opening meld has been postponed very late, it's probable that after the opening you'll be able to complete one canasta and go out very quickly. In such a case, you've seen which cards the opponents get and you can infer quite precisely which cards your partner has. Thus, your hands match probably very well in such a situation.

Opening and going out in the same turn gives you a 100 point bonus. Sometimes it's good strategy to try to do that - not because of the ridiculous 100 point bonus but because of the surprise effect.

Your main goal is going out. To achieve this goal, you may have to discard such cards that fit opponents' melds. Even though your opponents will get a few more points by taking them, you should give them, if it helps your own team go out. If you don't go out, you'll certainly lose much more.

A book I've read says that you might freeze the pile in order to safely get rid of cards that fit opponents' melds (or let your partner get rid of such cards) and then go out. Personally, I tend to disagree. When you are planning to go out, you need your wild cards to make a canasta or get rid of pairs. Usually you cannot afford a wild card for freezing the pile. Furthermore you don't know which cards you will get from the deck. Because you'll have to meld or discard all cards (except for one) in order to go out, you may have to give up even the frozen pile to the opponents in order to get rid of something you got from the deck after freezing the pile. (I really would like to hear other players' comments on this.)

Freezing the Pile

The main effect of freezing the pile is that after freezing you cannot take the pile with your melds. Because taking the pile becomes more difficult, frozen piles tend to become bigger and more valuable.

This means that if your opponents have a lot of melds that help them take discard piles, you can neutralize the effect of those melds by freezing the pile. Because a frozen pile grows big and valuable, your opponents can't ignore the frozen pile and keep on building canastas: If you freeze the pile, they must either go out quickly or start fighting over the pile. If either partnership eventually gets the frozen pile, that partnership becomes very dominant. Freezing the pile is a strategy that shakes the prevailing dominant-submissive setting; if you are yourselves dominant, you do not want the prevailing setting to shake.

When you fight over the pile, you mostly try to prevent your opponents from taking the pile: You take the pile when your opponents fail in defending it. The more cards you have, and especially the more safe discards you have, the better chances you have in fighting. Remember also that if your partner has only a few cards, he has little choice of discardable cards and he will eventually have to give the pile to opponents. Because one screw-up is enough to lose the battle, you should turn the game into a battle over the frozen pile only when both you and your partner are able to defend the pile.

So, the ideal situation for freezing the pile in order to start a battle is the following: You are submissive, and your opponents have a lot of melds, but no 6 card melds or canastas (if they have a canasta, they may go out and not fight). Also, they have less cards in their hands than you and your partner. You have both safe cards and pairs, preferably one surprising pair, and your partner has a many cards in his hand, but it seems that it's unlikely to get canastas and go out. In a really ideal situation your right-hand opponent has only a few cards (This means that it's not dangerous if your partner runs out of safe discards, because the right-hand opponent is unlikely to have pairs.) Also, the right-hand opponent is likely to run out of safe discards and you will have uses for your pairs.

Starting a battle over a frozen pile is a strategy I personally dislike. It stops strategical decisions and makes counting cards the most important skill. Often the pile grows and grows and the stakes grow and grow. If either side gets the pile, they often score several thousands of points (I've once got a huge discard pile worth over 5000 points). Furthermore, you must know that your partner is also counting cards; you can not defend the pile alone. If you can't choose your partner, you'd better try some less risky strategy.

Sometimes, however, an unfrozen pile grows so big that the team that takes it gets the dominance. In such situations the status quo is already broken and you may have to freeze the pile just because it increases your team's chances to defend and finally get it.

Dominant Play

If your partnership has many cards, it's probably easy to get even more by taking discard piles and make a lot of canastas - maybe even a couple of pure canastas. You try to make the hand last as long as possible so that you'll complete as many canastas as possible. It's not you but your opponents that try to go out. You should go on strong until they're able to go out, and just before that go out yourselves or at least meld everything valuable.

Sometimes both members of your partnership have a lot of cards, and the game is easy. Sometimes the cards keep on accumulating to one player, and the other player loses his cards by melding. If you have only one or two cards, you're extremely vulnerable because the choice of discards is very limited; you may have to give the discard pile to opponents. Thus, when your partner has a lot of cards but you do not, you must think really carefully before each melding; it's your partner's game. You just stand back and support him by having a hand that's strong enough. On the other hand, it'll be quite easy for you to go out. If you're running out of cards, it is sometimes a good idea to go out even though your partnership is dominant.

Sometimes, you may have completed some canastas and you have no (or little) promising 5 or 6 card melds left. Despite the submissive position, the opponents may have got a couple of 5-6 card melds. In such situations, consider carefully which team will get more points if the game goes on and go out if you think that your opponents will get more.

In general, the hand does not last as long as the dominant side thinks. The submissive side goes out sooner or later and it's then really embarrassing to find out that you are having cards in your hand that would have made a canasta if they were melded.

To Meld or not to Meld?

The main reason for melding is that you try to build a canasta from that meld. Thus, it's more efficient to make fewer but longer melds than to make very many 3 card melds. (When either team is very likely to go out, you may have to meld everything you can to maximize your score.)

The main reason for not melding is that it's easier to defend the discard pile if you have a lot of cards in your hand. If you have only a few cards in your hand, you cannot play if the opponents freeze the pile. You'll have no choice of discards and no pairs to take the pile. So, keep cards in your hand and meld only such combinations that can probably turn into canastas. On the other hand, it's generally advantageous to meld combinations that have a good chance to evolve into canastas: After the combination is melded, your partner knows that he must not discard the cards that fit the meld (unless he desperately needs a safe discard.)

In particular, avoid melding the same cards as your opponents. Such melds have little chance to evolve into canastas, because your opponents have quite a lot of the cards needed for the canasta. Also, if you meld them, the opponents will see that their meld has little chance to evolve into a canasta, and thus they have more information that they can use when they plan their strategies.

If you want to meld a new combination, meld just three cards even though you have more in your hand. This tactic has three advantages: (1) If the opponents freeze the pile, you may have a surprising extra pair in your hand (if it was originally a 5 card combination.) When the pile is frozen, people often discard cards that fit their opponents' melds. (2) The opponents may have to discard a card that fits your meld. Then they usually choose the card that fits the shortest meld. Because your 3 card meld is very short, they may give you the 5th or 6th card of that rank (count also those cards that you have in your hand.) When the discard pile is small, people sometimes discard cards that fit opponents' 3 card melds just because they want to save their good discards for bigger piles. (3)You're stronger when you have more cards in your hand.

Finally, of course, you probably will also meld the extra cards in your hand, but you should postpone it until the meld is close to a canasta (that is, until your partner continues the meld or you get more cards for that meld yourself) or someone is likely to go out.

If your partner usually melds only 3 cards from a 4 or 5 card combination, you usually should continue his melds if possible with natural cards. This way he gets to know how close to a canasta the meld is and it's easier to him to decide how to use the hidden resources. I've sometimes used such strategy that players add one card at a time in turns into a meld. This way the your partnership has quite a lot of information about the meld but the opponents do not. The strategy works fairly well if you're ready to deviate from it when needed; however, it's not The Golden Rule of effective melding.

Melding a wild card must be considered more carefully than melding a natural card: a natural card fits in only one meld. As long as a wild card is in your hand, it's very flexible: You can meld it almost anywhere. When you meld it, it loses its flexibility. Thus, apart from the opening meld and taking discard piles, do not meld wild cards in the beginning or in the middle of the game, unless you really need to make a canasta quickly. (And even then, make sure that melding the wild card really helps making the canasta.) Later you will see which natural cards your team has got and you can use the wild cards more efficiently.

It's very rare that you should meld a pair with a wild card. (The opening meld, taking the discard pile and going out are exceptions to this rule.) Almost always you just waste the wild card into a combination that has no chances to turn into a canasta, and you waste a pair that's useful if you have to fight over a frozen pile. One exception could be a case when you know that your partner also has a pair of the same rank. (He has, for example, taken a discard pile that contains the pair.) If you think that the partner is likely to discard the pair, you could save your common 4-card combination by melding your pair with a wild card.

In general, you should add natural cards to 4-6 card combinations if possible; your goal is making canastas. Adding wild cards must be considered more carefully, even though you could make a canasta by adding a wild card. If you make the canasta, the wild card will lose its flexibility, but there's also another point that you must take into account: When the opponents must discard cards that fit to your melds, it's best for them to discard such a card that fits to a complete canasta. This way they do not help you to get more canastas. Thus, each complete mixed canasta gives your opponents more good discardable cards, which is a bad thing for you. Furthermore, you'll get an extra 200 point bonus for each pure canasta. Eventually, before the hand ends, you must make canastas with your wild cards, but this should be postponed as late as possible. (On the other hand, postponing it too late has disasterous effects.)

Miscellaneous Advice

Team play

You play together with your partner. When you are making a decision of which strategy to use, you must take into account also your partner's strength. When it's obvious that your partner tries to do something (e.g. go out), you should in general help him do that. In the other hand, you should be helped by your partner when you try to do something (e.g. try to get a lot of canastas by making the game last longer.) When neither of you is egotistic and both players are helping each other, the results are good.

With clear and consistent play, it's usually possible to make your partner understand what you want to do. E.g. If you think the hand is over, and you should meld everything you can and then go out, tell it to your partner by making a couple of melds that do not help you get more canastas. (Add a wild card to a pure 3-4 card combination or make a meld of such cards that your opponents also have melded etc...) Remember that you are not allowed to convey information by speaking, expressing anger or joy, using secret signals (e.g. "foot signals" under the table), etc... In fact, even hesitating too long when you have a difficult decision to make is regarded as unethical information.

Opening Zones

Try always to calculate the scores of you and your opponents. This is because you should know if you or your opponents are going over an opening zone border. If you would end up with a score a little under 1500 or 3000, consider going out. If you go out, you'll have a better chance to get a good score in the next hand with a small opening. Correspondingly, if your opponents are ending up with a little under 1500 or 3000 points (and you're not), consider not going out. If you, by continuing, get more points, you gain. Your opponents will only lose if they get one more canasta; a lower requirement for the opening meld is certainly worth more than 300 points. If you have to break an opening zone border, do it with a good margin!!! When you calculate scores the teams will end up with, take into the account the 100 point bonus for going out.

Red threes

When your opponents have 3 red threes and you have none, they have a better chance to get more points with red threes. Both have equal chances to get the last red three, but it'll be worth 100 for you, and worth 500 points for your opponents. When the situation is otherwise even, this may be a good reason for you to go out.

Sometimes the opponents have not made the opening meld and they have some red 3's. In such situations, they'll get something like -500 - -300 points if you go out, but +300 - +500 points if they open. It can be a good idea to go out and take that 600-1000 point advantage rather than continue the hand and try to get a bigger but more uncertain victory.

"May I go out?"

You're allowed to ask your partner "may I go out". It's possible to use it as a signal: When you're able to go out and your partner has a lot of cards in his hand, you can tell him to get rid of his cards by asking "may I go out". He answers "No" and when it's his turn he melds everything he can. When it's your turn again, you go out. Note: When you use this strategy, you should not break your going-out -hand after the first negative answer. (Remember, that the rules require you to be able to go out when you ask "May I go out".)

Ask "may I go out" only when you want your partner to make the decision (the exception to this is when you're using the strategy described above). This way, when you ask, your partner knows that he's the only person having some important information, and this knowing-that-he-knows helps him to make a better decision. If you can make the decision by yourself, without consulting your partner, do not ask just because you want to be polite.

(Example: You have one canasta, one mixed five or six card combination melded and no other important melds. It's clear that the only reason for not going out is trying to make a mixed canasta out of that meld. If you can go out and make that canasta, do so and ask nothing. This way, if you ask "May I go out?", your partner can be sure that it's impossible for you to complete that canasta, and he can decide if it's possible for him to make the canasta.)