Splitting pairs: the complete guide
Four pairs have fixed rules and the rest depend on the dealer. This guide covers all of them, with the reasoning that makes each decision stick.
What splitting really costs and buys
Splitting turns one hand into two, and that framing hides the important part: it also turns one bet into two. Every split is a decision to put more money on the table, so the question is never just "are two hands better than one" but "are two hands better than one by enough to justify doubling my stake". Sometimes the answer is yes because you are attacking a weak dealer. Sometimes it is yes because your combined total is so bad that almost anything beats playing it straight. And sometimes both are true at once.
The two always rules
Always split aces. Played as one hand, a pair of aces is a soft 12, which is close to the worst total you can hold. Split, each ace becomes the start of a hand that turns into 21 nearly a third of the time. The one catch is that most casinos, including the rules assumed on this site, give each split ace exactly one card and no more. Even with that restriction, splitting aces is the single most profitable split in the game against every upcard.
Always split 8s. A pair of 8s is 16, the worst standing total in blackjack. It loses to every dealer hand that finishes, and it busts more often than not if you draw to it. Each 8 on its own is the start of a mediocre hand, but mediocre beats terrible: against strong dealer cards, splitting 8s does not turn a loser into a winner, it turns a big loser into a smaller one. Damage control is still correct play.
The two never rules
Never split 10s. Twenty is the second best total in the game and wins outright most of the time. Splitting it trades one near-certain winner for two merely good hands, and no upcard makes that trade profitable. You will occasionally see someone split 10s against a dealer 6 and win both hands. The rankings table shows why that is still a mistake: the expected value of standing on 20 beats the split against every upcard.
Never split 5s. A pair of 5s is 10, one of the best drawing totals you can hold and a doubling hand against most upcards. Two hands each starting with a 5 are two weak hands. Treat the pair as a 10 and it usually becomes a double down.
The judgment pairs
9,9 splits against 2 through 6 and against 8 and 9, but stands against a 7. The stand looks odd until you remember what a dealer 7 usually becomes: a 17. Your 18 already beats it, so there is nothing to fix. Against an 8 or 9 your 18 is in trouble, and splitting into two 9-start hands fights back better than standing.
7,7 and 6,6 split against the weak upcards, roughly 2 through 7 for the 7s and 2 through 6 for the 6s. The logic is offensive rather than defensive: 14 and 12 are poor totals, and against a dealer likely to bust it is worth paying for a second hand.
2,2 and 3,3 split against 2 through 7 when doubling after a split is allowed, as it is here. Much of their value comes from the chance that one of the new hands becomes a 9, 10 or 11 you can double.
4,4 is the rarest split: only against a 5 or 6, and only because doubling after the split is allowed. Otherwise a pair of 4s is just an 8, and you hit it.
Check any of this yourself
Every claim above is a number, not an opinion. Put any pair into the hand analyzer and compare the expected value of splitting against the alternatives, or find the full pattern at a glance in the pairs section of the strategy chart.