Splitting Pairs: Always, Never and "It Depends"

Splitting Pairs: The Three-Tier Framework

Key Takeaway: Always split Aces and 8s. Never split 10s or 5s. Every other pair depends on the dealer's upcard and on whether the house lets you Double After Split (DAS). DAS is the rule that moves 2s, 3s, 4s and 6s from "hit" to "split" against the dealer's weakest cards. Learn the three tiers and you remove the most common pair-splitting leak at the table.

Most players treat a pair as an invitation. It is not. A pair is a fork in the decision tree. You compare the expected value of playing the two cards as one hand against the expected value of two separate hands, each starting from a single card and each carrying a full bet. That comparison has a clean mathematical answer for every pair against every upcard. Your job is to memorize the answer, not to rediscover it with your gut every shoe.

I group the ten possible pairs into three tiers:

• Tier 1 (Always): pairs you split every time, no exceptions worth discussing at basic-strategy level.
• Tier 2 (Never): pairs you never split, whatever the dealer shows.
• Tier 3 (It Depends): pairs whose correct play turns on the upcard and the table rules.

If you have not yet locked down the full chart, start with our learn basic strategy guide and come back here. This article assumes you know your hard and soft totals cold and want the reasoning behind the pair matrix.

Tier 1: The Non-Negotiable "Always Split" Pairs

Aces: Turning a Liability into Two Premium Starts

A,A played as one hand is a soft 12. It cannot bust on the next card, but it is a weak total that usually needs two or more draws to become competitive. Averaged across dealer upcards it carries negative expectation, roughly −0.12 per unit under typical multi-deck rules. Split the pair and each hand starts with an 11-equivalent: an Ace that turns into 21 on any ten-value card and into a strong total on most other draws.

The usual restriction is that each split Ace receives exactly one card. Even so, the swing is large. Against the dealer's weak upcards, the pair's combined value climbs to roughly +0.50 or better per original unit. That is one of the largest single-decision EV gains in the game. Two facts drive it:

• Ten-density. Sixteen of every 52 cards are ten-valued, so each split Ace has about a 30.8% chance of becoming 21 immediately. Note that this 21 pays even money, not 3:2.
• Strong landing totals. A 7, 8 or 9 gives you soft 18, 19 or 20. That is another 23% of the deck that produces a standing total beating most dealer outcomes.

Against a dealer Ace or 10, splitting Aces still beats every alternative. Each split hand is worth less there, but soft 12 against a strong upcard is worse still. Split. Every time.

8s: Escaping the Worst Total in Blackjack

8,8 is hard 16, the worst total you can hold. You are too high to hit comfortably: 8 of 13 ranks bust you. You are also too low to stand, because 16 only wins when the dealer busts. Splitting gives you two hands that each start at 8. That starting point is mediocre, but it is far better than 16, and an 8 that catches a 2 or 3 becomes a doubling hand if DAS is allowed.

Against dealer 2 through 7, splitting 8s is offensive: each 8 is roughly break-even or better, and two such hands beat one doomed 16. Against 9, 10 and Ace it becomes defensive. Here are the numbers players argue about most, against a dealer 10 in a standard six-deck game:

Six cents on the dollar looks small. It is not. It is a structural edge the house collects from every player who flinches at "putting more money on a losing hand." The money you add is not new risk thrown at a bad spot. It replaces one hand with roughly −54% expectation by two hands with roughly −24% each. Splitting 8s against a 10 is damage control, and damage control is how disciplined players stay at the table.

The single refinement worth knowing: in some H17 games that offer late surrender, surrendering 8,8 against a dealer Ace edges out splitting by a hair. Without surrender, or in S17 games, split. For basic-strategy purposes, 8s are an always-split pair.

Tier 2: The Forbidden "Never Split" Pairs

10s and Face Cards: Do Not Break Up a 20

Hard 20 is the second-best hand you can be dealt, behind only a natural. Standing on 20 against dealer 2 through 9 carries expectation in the range of +0.65 to +0.75 per unit, and it stays strongly positive even against a dealer 10. You win when the dealer busts or finishes anywhere from 17 to 19. You push on 20. You lose only to 21.

Splitting 10s trades that near-guaranteed profit for two hands starting at 10. A 10 is a good starting card, but two of them are not worth more than a made 20. Against a dealer 6, for example, the split hands combined fall clearly short of standing. You have doubled your exposure to buy a lower expected return. That is a pure leak, and it is one of the clearest tells of an amateur. We break the error down further in our splitting tens mistake analysis.

There is one technical footnote. Card counters split 10s against a dealer 5 or 6 at high positive true counts, roughly +5 and +4 respectively in multi-deck games. At those counts the remaining shoe is so rich in tens that each split hand becomes worth more than half of a 20. That is an index play, not basic strategy. If you are not keeping an accurate running count and converting it to a true count, the play is simply "never." Doing it at a neutral count combines a huge EV loss with a pit-boss signal.

5s: A Hard 10 in Disguise

5,5 is hard 10, one of the best doubling totals in the game. Split it and you get two hands starting at 5. A 5 is a weak starting card: it often becomes a stiff total between 12 and 16 after one or two draws. Combining the two 5s into a 10 is worth far more than separating them.

Play 5,5 exactly as hard 10:

• Double down against dealer 2 through 9. About 31% of the time you catch a ten-value card for 20, and an Ace gives you 21.
• Hit against a dealer 10 or Ace. The dealer's strong upcard makes doubling unprofitable at a neutral count.

Never split 5s. There is no realistic rule set where splitting them is correct.

Tier 3: The Conditional "It Depends" Pairs

This is where most of the skill lives. The correct play changes with the dealer's upcard, and for several pairs it also changes with whether DAS is allowed.

2,2 and 3,3: Small Pairs That DAS Turns On

Without DAS, split 2s and 3s against dealer 4 through 7. With DAS, extend that to dealer 2 through 7. Hit against 8, 9, 10 and Ace.

A starting 2 or 3 is weak on its own. Its value comes from what it can become: a 2 or 3 that catches a 7, 8 or 9 turns into a 9, 10 or 11, a natural doubling hand against a weak dealer. When DAS is allowed, that second-stage double adds enough EV to push 2,2 and 3,3 against dealer 2 and 3 from "hit" into "split." Against a 7, splitting works because the dealer often ends on 17, and two hands that can each reach 18 or better are worth more than one hand of 4 or 6.

4,4: The Pair That Exists Only Under DAS

With no DAS, never split 4s. Hard 8 is better played as one hand: hit it. With DAS, split 4s against dealer 5 and 6 only. Each 4 can then catch a 6 or 7 and become a doubling 10 or 11 against the dealer's weakest cards. It is a small edge, but it is real.

6,6: Split Against Weakness

Without DAS, split 6s against dealer 3 through 6. With DAS, add dealer 2. Hit against 7 and above.

Hard 12 is a stiff total. Two hands starting at 6 are not strong either, but against a dealer 3, 4, 5 or 6 the dealer busts often enough that splitting earns a positive return. Against a 7 or higher, the dealer is too likely to make a strong hand, and splitting just puts more money at risk in a losing position.

7,7: Split Through 7, Hit Above It

Split 7s against dealer 2 through 7. Hit against 8, 9, 10 and Ace. A few notes:

• Hard 14 is a stiff total. Against weak dealer upcards, two hands starting at 7 are worth more than one 14.
• Against a dealer 7, each split hand only needs 18 to beat the dealer's most common result, 17. Two chances at that are worth more than standing on 14.
• Against 8 or higher, splitting fails. Each 7 needs at least two draws to reach a competitive total, and the dealer's upcard is too strong. Hit, or surrender against a 10 if late surrender is offered and the rules call for it.

9,9: Why You Stand Against a 7

Split 9s against dealer 2 through 6, 8 and 9. Stand against 7, 10 and Ace.

The 7 exception is where intermediate players get caught. They see "split vs 2–6 and 8–9" and assume the 7 gap is a misprint. It is not. Against a dealer 7, your 18 already beats the dealer's most likely final total. Roughly 37% of the time a dealer 7 finishes at exactly 17, and roughly 26% of the time it busts. Standing on 18 wins outright in about 63% of outcomes. Splitting would trade that made hand for two 9s that must each draw to something better. Standing is worth roughly +0.40 per unit, and splitting falls short of that.

Against 8 and 9 the logic reverses. The dealer is likely to finish at 18 or 19. Your 18 is at best a push and often a loser, so two hands starting at 9 are worth more. Against 10 and Ace, the dealer is strong enough that neither play is good, and standing on 18 loses less than splitting.

Rule sets differ from casino to casino. Use the Strategy Chart Builder to generate the exact pair matrix for your game: deck count, S17 or H17, DAS, surrender and resplit limits.

How DAS Flips the Borderline Pairs

Double After Split is worth roughly +0.14% to the player in a typical six-deck game. Most of that value comes from the conditional pairs. The mechanism is simple: when you split a small pair, a meaningful share of the resulting hands land on 9, 10 or 11, the totals where doubling against a weak dealer is most profitable. Without DAS, those hands can only be hit. With DAS, you collect the full doubling premium on each one.

That extra value is what pushes 2,2 and 3,3 vs dealer 2–3, 4,4 vs 5–6 and 6,6 vs 2 from hitting to splitting. If the table forbids DAS, those additions disappear, and you fall back to the narrower no-DAS ranges in the table above. Our double and split restrictions guide covers how other restrictions, such as limits on resplitting and doubling only on 10 or 11, shift these decisions further.

Resplitting Aces: A Small but Real Edge

Many games let you split Aces only once. Some allow Resplitting Aces (RSA): if a split Ace catches another Ace, you can split again rather than being stuck with a soft 12 that cannot take another card. RSA is worth about +0.08% to the player in a typical six-deck game. The frequency math shows why the edge is small but worth seeking out:

For a full combinatorial treatment of split sequences, including resplit trees and the card-removal effects that shift these frequencies in single- and double-deck games, see the pair splitting combinatorics on BlackjackMath.org.

The Most Common Pair-Splitting Leaks

These are the errors I see most often from intermediate players, ranked by how much they cost:

1. Splitting 10s. The most expensive habit on this list. You break up a hand worth roughly +0.65 to +0.75 to chase two hands worth less in total.
2. Refusing to split 8s against a 10 or Ace. Fear of "throwing good money after bad" costs about 0.06 units every time this spot comes up.
3. Splitting 9s against a 7. Your 18 already beats the dealer's most likely 17. Stand.
4. Splitting 5s. Treat 5,5 as hard 10 and double against 2 through 9.
5. Using DAS ranges at a no-DAS table, or the reverse. Check the rules and adjust your 2s, 3s, 4s and 6s before you play.

Drilling the Three Tiers

Memorize Tier 1 and Tier 2 first. Four pairs, two rules, no exceptions at basic-strategy level. Then work through Tier 3 one pair at a time, starting with 9s, because the 7 exception is the most counterintuitive. Next learn 7s (split through 7), then 6s, then the small pairs. Drill each pair against all ten upcards until the answer arrives without hesitation, then run mixed drills that combine pairs with hard and soft totals. At a live table you get no time to look things up, and a split decision you have to think about is one you will eventually get wrong.

Pair splitting is the most mechanical part of basic strategy and one of the easiest places to recover lost expectation. The math does not care how many chips are on the table or how bad the last shoe was. Split Aces and 8s, keep your 10s and 5s together, and let the upcard and the rules decide the rest.