Every turn you have to throw a card away. Nine times out of ten, that card should be the highest one in your hand that isn't doing anything for you. It's the plainest discard rule in Rummy. It's also the one beginners most often break, and it's the one that quietly halves your losing scores.
Round twelve of a Sunday afternoon session. You've been holding an offsuit King all game. It's not doing anything for you. It doesn't pair with any of your other cards, it doesn't sit near a run, it just floats there. You keep telling yourself, "the moment another King comes, I'll have a set."
The other King never comes. Your opponent declares. You flip your hand over. You still have that King. Ten points, straight to the loss column. On top of the six or seven other unmatched cards you couldn't finish.
This is the single most common way beginners lose points in Rummy. Not by playing bad melds. Not by missing sequences. By holding onto high-value cards that never became anything, hoping they would.
The rule that fixes this is one sentence long. Every turn, discard your highest useless card first. Not the middle one. Not the one you're least attached to. The highest one that isn't part of a live meld or near-meld. Just throw it.
This guide unpacks that rule. Why it works. When it's obviously right. When it feels wrong but is still right. And when to actually bend it. If you play on Play Rummy, the app will suggest discards for you, but the rule underneath that suggestion is exactly this.
Before we go further, let's make sure we're on the same page about what "high" means in Rummy scoring.
In most variants of Play Rummy, the point values are:
At the end of the round, if you haven't declared, you add up the points of every card in your hand that isn't part of a valid meld. Those are your deadwood points. The lower that number, the less you lose by.
So a King, a Queen, and a Jack in hand at the end is 30 points of deadwood, all by themselves. Compare that to a 2, a 3, and a 4 of mismatched suits: only 9 points. Same number of unmatched cards. Vastly different scores.
Every high card you keep is a bet. You're betting the round ends in your favor before it costs you. And most rounds do not end in your favor. Someone else declares. So the bet loses more often than it wins.
Three reasons the highest-useless-first rule beats every alternative for beginners and intermediates.
You don't know when the round will end. On any given turn, from turn three onward, someone could declare. When they do, whatever is in your hand becomes deadwood. High cards make that deadwood expensive. The only way to protect yourself from a sudden declaration is to have already shed your heaviest useless cards.
High cards are harder to complete. There are fewer ways to build a meld around a Queen than around a 6. A Queen can only sit in Q-K-A? No, Ace is low, so Q-K is the ceiling and you need J-Q-K or 10-J-Q. Two sequences total. Or a set of Queens. Compare that to a 6, which can sit in 4-5-6, 5-6-7, or 6-7-8, plus a set of 6s. Twice as many paths. High cards are simply less flexible.
Discard piles love low cards. Your opponents are also trying to shed. That means they're throwing their own low, useless cards more often. High cards get held. So even if you're waiting on a matching Queen to build a set, that Queen is unlikely to hit the discard pile. It's more likely to be stuck in someone's hand or the stock. Waiting for high pairs is a slow lane.
Add these three together and you get a clear principle. High cards cost more if you get caught with them, complete less often when you try to build with them, and get released into the game less often when you're hoping to catch one. Three reasons to throw them first.
The trick to using this rule is speed. If you have to think for thirty seconds every turn about which card is truly useless, the rule turns into a chore. Here's the fast version.
Sort your hand by suit and rank. Now look at each card and ask two questions:
If the answer to both is no, the card is useless. It's a singleton. It has no partners in your hand. That's the definition. Nothing dramatic about it.
Now look at all your useless cards. Pick the one with the highest point value. Throw that one.
Look at that hand. The 4-5-6♠ is a sequence, safe. The 9-10♥ is a near-sequence, keep. That leaves the 7♣ and Q♦ as singletons. Both useless. The Q♦ is worth 10, the 7♣ is worth 7. Q♦ goes first.
This is the entire beginner version of the rule. Two questions per card. Highest useless goes.
Here's the trap every beginner falls into. You have a floating King. You know the rule says to throw it. But you think: "if I hold it one more turn, maybe I'll draw another King and I'll have a pair. Just one more turn."
One more turn becomes two. Two becomes five. Five becomes the whole round. The King never pairs. The round ends. Ten points, gone.
Here's the math you need to see. There are three other Kings in the deck (or in a two-deck game, seven). Roughly a third of them are in opponent hands. The rest are split between the stock pile and the discard pile. On any single turn, the odds that you draw a King are maybe one in twenty. Even if you're actively looking for one, you'd expect to wait ten to fifteen turns for it, on average. That's most of the round.
Meanwhile, every turn you hold that King, you're exposed. If the round ends this turn, that's 10 points. Next turn, 10 points again. And again. The cost isn't paid all at once. It's paid every single turn you keep it.
The rule doesn't say never hunt for a set of Kings. It says don't hunt for one just because you already have a floating King. That's the "sunk cost" mistake in card game form. The King in your hand right now is deadweight until it has a partner. Treat it like deadweight.
Once you've internalized the rule, you'll notice there are situations where blindly throwing your highest useless card isn't the best move. Intermediate play is about knowing the exceptions.
When the highest useless card is safe and the second-highest is dangerous. Say you have a lone Queen of Clubs and a lone 8 of Hearts. The 8♥ is lower, so the rule says throw the Queen. But your opponent has been picking up hearts all round. Throwing the 8♥ would probably feed them a card that completes a run. Throwing the Q♣ is safer, even though it's worth more. The rule shifts: highest useless, subject to safety.
When you're one card from declaration. If you have every meld ready except one final card, and you're hoping to draw it, sometimes you keep a slightly heavier card because it could be that final card. But only "one card from declaration." Not "three cards away, but I really believe in this Queen." Believing isn't a strategy.
When the highest useless card is a wild joker in a variant that allows it. If jokers are wild, a "useless" high card might not be so useless. It could be a joker substitute if the rules count face value. Check the variant. Play Rummy shows joker rules on the game screen, so you're never guessing.
When the round is about to be drawn. In some variants there's a fixed number of stock cards. When the stock runs low, the round will end by count instead of by declaration. If you can see that ending coming in two turns, sometimes it's worth holding a Queen for one more turn if it might still complete. But usually, no. Even in this case, if the Queen has no path to a meld, throw it.
A more advanced version of the rule uses the discard pile as a source of information.
Every card in the discard pile is a card that someone chose to throw. If you see two Kings hit the discard pile in the first four turns, that tells you something. It tells you that other players don't have King pairs. It also tells you that half the Kings in the deck are already out. Which means your own lone King has an even smaller chance of pairing.
The rule now has extra force. If your useless card's matching partners are already in the discard pile, it's not just useless, it's dead. It cannot ever become useful. Throw it immediately, no matter what else is in your hand.
The intermediate move is to keep a rough mental tally of what's in the discard. You don't need to memorize every card. You just need to notice when the partners of your held-onto singletons show up. When they do, drop your singleton immediately.
At the expert level, your discards are messages. Your opponents are reading them. The rule "highest useless first" is a great baseline, but it's also predictable. And predictable can be exploited.
Here's what an expert opponent reads when you throw a Queen. They think: "That player has no matching Queens, no Queen-adjacent runs (like J-Q or Q-K in that suit), and no plan for that card. So they're probably not building in the high end of that suit." They now have a small piece of information about your hand. Multiply this across turns and they can guess your suit strengths.
Experts occasionally disguise their discards. They might throw a middle card instead of the strict highest useless, if it doesn't cost them much and it hides information. They might discard slightly asymmetrically across rounds to break the pattern.
But even at the expert level, the rule still governs 80 to 90 percent of discards. The disguised discards are the exception, not the norm. If an expert threw random cards to confuse opponents, they'd cost themselves points on average. The rule wins in the long run.
The expert layer is really about knowing when disguise is worth the cost. Rarely against a bad opponent. Sometimes against a great one. Almost never as a habit.
One more expert layer. Experienced players don't just plan the next discard. They plan the next two.
Here's why. Every discard changes what your hand looks like on the turn after. If you have a Queen and a Jack, both useless, both singletons, throwing the Queen means the Jack becomes your highest useless next turn. You've committed to throwing the Jack unless something changes.
Expert players look at their hand and mentally chain two or three discards. "Queen this turn. Jack next turn. Then I'll draw the 8 I need." That chain lets them see when a plan is coming together, or when they're about to strip their hand down to nothing but low cards with no path forward.
The two-turn plan also protects against surprise. If your first discard opens up a card that's now your new highest useless, and that new card is a high-value one you didn't notice was floating, you can adjust. Better to see it coming than to be surprised by your own hand.
This all sounds heavy. In practice it's a two-second glance at what your hand will look like after this turn's discard. That's it. The two-turn plan is just a habit of asking "what's next?" before you throw.
A few discard habits that quietly cost beginners points, over and over.
Keeping face cards "just in case." Kings, Queens, and Jacks feel important because they're worth ten. That's exactly why you shouldn't keep them without a plan. Their weight is a cost, not a bonus.
Throwing low cards to feel productive. Some players discard whichever card is easiest to let go of, usually low ones. But a 3 of Spades that partners with a 2 and 4 you don't yet have is still worth 3 points, and it's occupying a slot. If it's not near anything, it's useless too, and it should go before the higher useless cards? No. Higher useless goes first. Always. That's the whole rule. If both a 3 and a Queen are useless, the Queen goes.
Confusing "useful" with "attached." Sometimes a card is emotionally attached (you've been carrying it since turn one, you feel invested) but is actually useless. Attachment doesn't count. Only mechanical usefulness (part of a meld or near-meld) counts.
Throwing a card that helps the opponent, just to obey the rule. The rule is highest useless first, subject to safety. Feeding an opponent a Queen that completes their K-Q-J run is worse than holding it one more turn. Read the table. If your opponent is clearly building in a certain suit, don't hand them fuel just because it's "the rule."
Here's a drill that welds the rule into your brain in about a week.
Play ten rounds against yourself or a friend. After every discard you make, say out loud (or write down) which card you threw and why. If your reason is "highest useless," great. If your reason is "I didn't like it" or "I was hoping for a pair," you've broken the rule. Note it.
At the end of the ten rounds, count how many times you broke the rule and what those breaks cost you at showdown. Compare that to how often the rule saved you points.
Do this drill for two sessions. By session three, the rule will feel natural. By session five, you'll be applying it without thinking. By session ten, you'll be one of the players who casually shrugs when someone else declares, because your deadwood is single digits.
Discard your highest useless card first, every turn. One sentence. One habit. It's the difference between losing rounds by 60 points and losing them by 15. Over a session, over a year, that gap is enormous.
Beginners follow the rule blindly and win more. Intermediates learn to bend it for safety and information. Experts know when to disguise their discards and how to plan two turns ahead. All three levels agree: the baseline is throw the biggest useless piece of weight in your hand.
If you play on Play Rummy, the game already highlights your deadwood so you can see the cost in real time. Use that. Watch how your deadwood number drops when you obey the rule and climbs when you break it. That feedback is the fastest teacher.
Every other tip about discarding sits on top of this one. Get this habit first. Everything else builds from here.