Why You Always Split Aces and 8s in Every Version of Blackjack

Splitting aces and 8s is the most durable rule in blackjack basic strategy. No dealer upcard changes it. No table variant removes it. No streak or session result overrides it. These two pairs are the only unconditional splits in the game, and the math behind each one is straightforward once you see the numbers.

Players who hesitate on these splits are paying a real cost every session. A player who refuses to split 8s against a dealer 10 is making the same mistake repeatedly, handing back equity on one of the most recoverable hands in the deck. This post explains exactly why both splits are mandatory and why no exception exists.

Why You Always Split Aces in Blackjack

A pair of aces counts as either 2 or 12. Both totals are weak starting positions. A hard 2 requires multiple hits before it becomes competitive. A hard 12 busts on any 10-value card, which represents roughly 31% of the remaining deck in most shoe games.

Splitting creates two separate hands, each starting with the strongest single card in the deck. An ace as a starting card gives you a 31% chance of receiving a 10-value second card and reaching a total of 21. No other starting card produces that probability.

The expected value comparison is decisive. Playing a pair of aces as a single hand produces a negative or near-zero EV depending on dealer upcard. Splitting aces generates a positive EV against every dealer upcard in the standard 6-deck shoe game. The split is not a close call on any row of the chart.

Why You Always Split 8s, Even Against a Dealer 10?

Hard 16 is the worst hand in blackjack. It is too high to hit without a serious bust risk and too low to stand with confidence against any dealer upcard above 6. A pair of 8s delivers exactly this: a guaranteed hard 16 before you take any action.

Splitting converts that single hard 16 into two hands each starting at 8. A starting total of 8 can reach 18 with a single 10-value card, which is a competitive total against most dealer upcards. The transformation from one losing position to two playable positions is the entire argument for the split.

Against a dealer 10, splitting 8s still produces a negative expected value. That is not a reason to keep the hand. The question is never whether a split produces a winning EV in isolation.

The question is whether splitting produces a better EV than the alternative. Against a dealer 10, hard 16 has an EV of approximately -0.54. Two hands starting at 8 against a dealer 10 produce an EV of approximately -0.38. Splitting is the smaller loss, and smaller losses compound favorably over a session.

Why Does the EV Math Compare Across All Dealer Upcards?

The expected value advantage of splitting aces is largest against dealer bust cards and remains strongly positive against dealer 10 and ace. No upcard produces a scenario where holding the pair of aces outperforms the split. The rule is unconditional because the math is consistent across every column of the blackjack strategy chart.

For 8s the picture is different but the conclusion is the same. Against dealer 2 through 7, splitting 8s produces a positive or near-neutral EV. Against dealer 8 through ace, splitting produces a negative EV. But in every case, the EV of the split exceeds the EV of keeping hard 16. That is the standard for an unconditional split.

Players who surrender 8s against dealer 9, 10, or ace are applying the right instinct in the wrong direction. If surrender is available and the rules permit it, surrendering hard 16 against dealer 9, 10, or ace is actually correct.

But when surrender is not available, the split is mandatory. Keeping the hand is the worst available option in that scenario.

What Is the Result?

Most licensed casinos allow only one additional card per split ace. You cannot hit further, double down, or re-split if you receive another ace. This restriction reduces the EV advantage of splitting aces, but it does not eliminate it.

With one card allowed per split ace, the probability of receiving a 10-value card and reaching 21 remains at roughly 30.8%. That outcome alone generates enough EV to make the split clearly correct. The one-card restriction costs you flexibility on the rare occasions when a non-10 lands, but the base case outcome is unchanged.

Note also that a 21 formed by a split ace and a 10-value card is not a natural blackjack. It pays 1:1, not 3:2. This is standard across virtually all casinos and is already factored into the EV analysis supporting the split. The restriction does not change the correct play.

If you want to see how these splits play out in real hands before committing to them at a live table, a real-money live dealer session is the fastest training environment available.

Playing real-money live blackjack at a qualified casino puts actual stakes behind every decision, which forces faster memorization than free play. Set a firm session limit before you start.

Why Players Resist These Splits

The resistance to splitting 8s against a dealer 10 comes from loss aversion. Players see a strong dealer upcard and do not want to put a second bet at risk.

The instinct is logical in isolation but wrong in application. The goal is not to minimize the bet in play. The goal is to minimize the expected loss across the hand.

Resistance to splitting aces is rarer but does occur. Some players believe that keeping a soft 12 gives them flexibility to hit toward any total. That flexibility is largely illusory. A soft 12 that receives a low card becomes a slightly better soft total. An ace that receives a 10 becomes 21. The comparison is not close.

Both resistances share the same source: outcome thinking rather than EV thinking. A player who splits 8s against a dealer 10, loses both hands, and concludes the split was wrong has made a logical error. The correctness of a decision is determined by EV at the moment of the decision, not by the result of a single instance.

Before you test these plays at a real table, run them through our free blackjack simulator practice unlimited hands at zero cost until every move becomes automatic.