You are holding two 10s. The dealer shows a 6, the worst upcard on the table. Somewhere in the back of your head a voice says: the dealer is going to bust, so why take one 20 when you could have two?
It is the most seductive wrong play in blackjack, because unlike most mistakes it is not obviously reckless. It sounds like advanced thinking. It is not, and understanding exactly why is more useful than memorizing the rule, because the same logic explains several other decisions on the chart.
What 20 is actually worth
Start with the hand you already have. A pair of 10s is a hard 20. Against any dealer upcard it is a winning hand more often than not, and against the weak upcards it is a strongly winning hand. Only a dealer 21 beats it and only a dealer 20 ties it, and both of those require the dealer to make a specific total from a bad starting position.
That matters because expected value is always measured against the alternative, not against zero. When people evaluate a split they picture the upside of two good hands and forget that they are giving up something very valuable to get there. You are not trading a mediocre hand for two chances. You are trading a near-best hand for two hands that each start at 10 and have to be built.
The comparison, in numbers
Take the friendliest possible case for the split: six decks, dealer stands on soft 17, dealer showing a 6. Per the commonly published expected value tables, standing on 20 in that spot returns about 0.70 units per unit bet. Splitting returns about 0.62.
| Play, 10s versus dealer 6 | Approximate return per unit bet | Money at risk |
|---|---|---|
| Stand on 20 | About 0.70 | One bet |
| Split into two hands | About 0.62 | Two bets |
Read that carefully, because the shape of it is the whole lesson. Splitting is profitable. In isolation it makes money, which is why it does not feel like a blunder when you do it and one of the hands makes 19. It is simply less profitable than standing, by roughly 8 percentage points of your bet in the single best spot for it, and it doubles the amount of money exposed to reach that worse result.
Against stronger upcards the gap widens sharply. Against a dealer 10 or ace, splitting a 20 is not close to correct under any circumstances.
The positive EV but worse EV trap
This is the pattern worth carrying away. A play can be profitable and still be a mistake, because blackjack decisions are ranked, not sorted into good and bad. The question at every decision point is never "does this play make money" but "does this play make the most money available here".
The same trap catches other hands. Doubling on a soft 18 against a 2 wins often enough to feel right, yet the published charts rank standing higher when the dealer stands on soft 17. Standing on a stiff hand against a small upcard wins a decent share of the time, but that is not the comparison that decides it. Learning to ask the ranked question rather than the binary one is most of what separates a chart follower from someone who understands basic strategy.
It also explains why expected value is the only sane unit for these arguments. Intuition reports how a play feels when it works. Expected value reports what it is worth across every way it can go.
21 TrainerFree basic strategy training with the reasoning behind every split, double and stand.
Get the app
Why the weak upcard argues for standing
The tempting version of the argument runs: the dealer is weak, so this is the moment to press the advantage. The flaw is that the dealer's weakness is already priced into your 20.
When the dealer shows a 5 or 6, dealer bust frequency is at its highest, in the low 40s percent per the commonly published figures. That means your 20 is winning a very large share of the time as it stands. The split does not get to compare itself against a neutral baseline. It has to beat a baseline that the dealer's weak card just pushed near the top of the range.
And the two new hands are not miniature 20s. Each starts with a single 10 and draws one card. Most of the time the result is a total in the middle teens or a soft hand, not another 20. You gave up a near-certain strong position for two average ones. The dealer's weakness helps those hands, yes, and it helps the hand you already had more.
Two more reasons in practice
Beyond the expected value, two practical points push the same way.
- Risk. Splitting commits a second bet equal to the first. In terms of bankroll and swing, splitting 10s is a bet size increase disguised as a strategy decision, taken at a moment when you already held the strong hand.
- Attention. Splitting 10s is unusual enough that it reads as unusual to everyone watching. That does not matter at a kitchen table. It matters to anyone practicing habits they intend to carry into a live game.
Compare this with the pairs you always do split. Aces and 8s are split because the starting hand is genuinely poor: 16 is the worst total in blackjack, and two soft hands starting from an ace are worth far more than a single soft 12. That contrast is the point. Splitting aces and 8s improves a bad hand. Splitting 10s damages a great one.
The counting exception, clearly labeled
There is one context in which splitting 10s is correct, and it is not basic strategy. Per Schlesinger's published list, the Illustrious 18 includes splitting 10s against a dealer 5 at a true count of about +5, and against a dealer 6 at about +4.
Those are index deviations: plays that override the chart only when the count says the remaining deck is unusually rich in tens, which raises the chance that each split hand becomes another 20. They apply to a counter tracking a true count, they require the actual count to be that high, and they are not a license to split 10s because the last three hands went well.
They also carry a practical caveat that experienced counters take seriously. Splitting 10s is one of the most conspicuous plays at a table. Many counters skip these two indices deliberately as cover, judging the attention they attract to be worth more than the small gain. If you are working through index deviations, learn them, and understand that appearing on the list does not mean playing them is obligatory.
How to drill the pair decisions
Pairs are a minority of the hands you get dealt, which means random practice gives you few repetitions of them and errors hide easily. Treat pairs as their own drill.
Work through all ten pair types against all ten upcards until each answer arrives without reasoning. Pay extra attention to the ones that shift with rules, such as splitting 4s where double after split is offered, and to the ones that feel wrong the way 10s do. Our guide to splitting pairs lays out the full set. In 21 Trainer, the free tier deals pair scenarios with a plain English explanation for each decision, and Pro's heat maps show whether pairs are the category quietly dragging your accuracy down.
As always, none of this makes an individual hand predictable. Correct play lowers the house edge; it does not remove it, and you will split aces perfectly and lose both hands more often than feels fair. The point of the chart is that it is right across thousands of hands, not that it rescues any one of them.
Frequently asked questions
Should you ever split 10s in blackjack?
Not under basic strategy. A pair of 10s is a 20, which is the second best hand in the game, and standing beats splitting against every dealer upcard. Splitting 10s is profitable in isolation against a weak upcard, it is simply less profitable than standing, and it puts a second bet at risk to get there. Basic strategy says stand every time.
How much does splitting 10s against a 6 actually cost?
Per the commonly published expected value tables for six decks with the dealer standing on soft 17, standing on 20 against a dealer 6 returns about 0.70 units per unit bet, while splitting returns about 0.62. Standing is worth roughly 8 percentage points more in that spot, and it risks one bet instead of two. Both plays are profitable, which is exactly what makes the split tempting.
Why does a weak dealer upcard not make splitting 10s better?
Because a weak upcard raises the value of both plays at once. When the dealer shows a 5 or 6, your 20 is already winning most of the time, so the standing baseline is very high. The split has to beat that high baseline with two hands that each start at 10 and will usually finish below 20. A weak upcard flatters the split, but it flatters standing more.
Do card counters ever split 10s?
Some do. Per Schlesinger's published list, the Illustrious 18 includes splitting 10s against a dealer 5 at a true count of about +5 and against a dealer 6 at about +4. This is an advanced count deviation, not basic strategy, it only applies when a rich deck makes tens likely, and it is a well known cover problem because splitting 10s draws attention from the pit faster than almost any other play.
What is the best way to drill pair splitting decisions?
Practice pairs as their own category rather than mixed into random hands, because pairs are a minority of deals and random practice gives you too few repetitions. Run through every pair against every upcard until the answers are automatic, and pay particular attention to the pairs that change with rules such as double after split. 21 Trainer deals scenarios with an explanation for each decision, so a wrong split teaches the reason.