When to Split Pairs in Blackjack

How splitting actually works

When your first two cards are a matched pair, you can split them into two separate hands. You place a second wager equal to your original bet, the dealer separates the cards, deals a second card to the first hand, and you play it out — hit, stand, sometimes double. Then the dealer moves to the second hand and you do it again. You're now playing two independent hands against the dealer's single hand.

Card rank, not suit, defines a pair for splitting. Any two ten-valued cards count as a pair: 10-J, Q-K, J-K all qualify. Most casinos allow up to three splits (giving you four hands total); some allow more. Aces are usually restricted: split aces typically receive one card each and cannot be hit further, and a 10 on a split ace pays even money as 21, not 3:2 as a blackjack.

The "always split" pairs

Always split A,A

Two aces total either 2 or 12 — both garbage. As a single hand you're hitting a stiff. Split them and each ace becomes the foundation of a hand starting at 11. About 31% of the time, your next card is a ten-value, giving you 21 on that hand. Even with the no-hit-after-split restriction, the EV of splitting aces is so strong that you do it against every single dealer upcard, including a 10 or an Ace.

The Wizard of Odds simulators show that splitting aces is worth roughly +0.18 units per bet across all dealer upcards on average. Standing on 12 or hitting aces as a soft 12 are both deeply negative.

Always split 8,8

A pair of 8s totals 16, the worst hand in blackjack. Against any dealer upcard 16 loses more often than it wins; against dealer 9, 10, or A it's a near-disaster. Splitting turns one terrible hand into two hands starting at 8 — and 8 is a workable starting total. A 10 on the next card gives you 18, a 7 gives you 15 which is still hittable.

The key insight: splitting 8s isn't always profitable in absolute terms (against a dealer 10 you still lose money on average) — it just loses less than standing or hitting on 16. If the table offers surrender, 8,8 vs 9, 10, or A becomes a surrender in some H17 rule sets, but in straight S17 with surrender unavailable, always split.

The "never split" pairs

Never split 5,5

5,5 is hard 10. Hard 10 is the second-best starting total in the game. Splitting it gives you two hands starting at 5, both of which are weaker than your original 10. Treat 5,5 as hard 10: double against dealer 2-9, hit against 10 or A. The only reason a player ever splits 5s is because they're treating "pair = split" as a reflex. Don't.

Never split 10,10

20 wins about 85% of the time against any dealer upcard. Splitting tens trades a near-certain win for two hands starting at 10 — which are good but nowhere near as good as 20. Card counters split tens at very high true counts because the next card is overwhelmingly likely to be another ten, turning one 20 into two 20s. For basic strategy with no count information, you stand on every pair of tens against every dealer card.

Never split 4,4 (with one DAS exception)

4,4 is hard 8. Hard 8 isn't great but it's hittable. Splitting 4s gives you two hands starting at 4, which is materially worse. The one exception: under DAS, splitting 4s against a dealer 5 or 6 becomes marginally correct because a 6 or 7 on the split hand lets you double for value. Without DAS, hit every time.

The conditional pairs

P* means split if DAS is allowed, otherwise hit. The chart assumes S17, late surrender unavailable.

9,9 — the most-missed pair

Split 9,9 against dealer 2, 3, 4, 5, 6, 8, and 9. Stand on 7, 10, and A. The 7 exception trips up almost everyone. Logic: 18 beats the dealer's most likely total when the upcard is 7 — the dealer makes 17 about 37% of the time on a 7 upcard, and you beat that. Splitting trades a winning 18 for two hands starting at 9 against a dealer 7, which is worse on average. Against 10 or A, the dealer's expected total is too high; you stand on 18 and hope for the best.

7,7 and 6,6

Split 7,7 against dealer 2 through 7 — you're trying to escape a hard 14 against weak-to-medium dealer cards. Against 8 or higher you just hit, because the dealer's strong upcard means two hands starting at 7 perform worse than one hand hitting 14.

Split 6,6 against 2-6 with DAS, or 3-6 without DAS. Hard 12 is awful, and two hands starting at 6 against a bust-prone dealer is materially better, especially when you can double if the next card is a 4 or 5.

2,2 and 3,3

Both follow the same pattern: split vs dealer 2-7 with DAS, vs 4-7 without DAS. The DAS expansion exists because each split hand frequently lands on 8, 9, 10, or 11 — exactly the totals where doubling pays. Without DAS, splitting 2s against a dealer 2 loses value.

How DAS changes the chart

DAS — Double After Split — turns six pair decisions more aggressive: 2,2 vs 2-3, 3,3 vs 2-3, 4,4 vs 5-6, and 6,6 vs 2. Each of those splits gains value because the resulting hands can be doubled when they land on 9-11 totals. Wizard of Odds quantifies the player-side gain from DAS at roughly +0.14% across a full session. If you're playing a no-DAS game, follow the no-DAS column and don't split those marginal pairs.

Other rule wrinkles

Resplit Aces. Most casinos don't let you resplit aces. If you split A,A and the next card is another Ace, you're stuck with two hands of soft 12. A few games (Vegas Strip "Liberal" rules) allow resplit aces and add about 0.08% in player EV.

Max splits. The standard cap is three splits / four hands. Some games cap at one split. The strategy doesn't really change — you split when basic strategy says to and stop when the casino makes you stop — but it does cap your upside on streaks of identical cards.

ENHC impact on pair plays. Under ENHC, splitting against a dealer 10 or Ace risks losing the doubled bankroll to a hidden blackjack. Strategy charts under ENHC drop a few splits against 10 (notably 8,8 vs 10 sometimes becomes a hit or surrender depending on rules). Always check the chart for the specific rule set.

The most expensive splitting mistakes

Splitting 10,10. Costs roughly 5% EV every time. You're trading the second-best hand in blackjack for two mediocre starts. Even at the most aggressive table conditions, only counters at extreme positive counts split tens.

Not splitting 9,9 against 9. Players see dealer 9, look at their pat 18, and stand. 18 vs 9 is a marginal loser; two hands starting at 9 vs 9 is a marginal winner. The swing is about 4% in EV.

Splitting 4,4 without DAS. No DAS means each 4 becomes a hand starting at 4 with no upside. Just hit 4,4 as a hard 8.

Splitting 5,5. The single worst splitting decision in the game. Treat 5,5 as 10 and double against 2-9.