Beginners look at their own two cards. Everyone else looks at the dealer's one card first. That habit is the single largest difference between a player who has internalized basic strategy and one who is guessing, because the dealer's upcard is what turns a total into a decision. A hard 12 is not a hand you play one way. It is a hand you play five different ways depending on what is showing across the table.
This is the logic underneath the whole chart, and it is worth understanding rather than memorizing, because understanding it makes the chart mostly self-explaining.
The dealer has no choices, which is what makes this possible
You can predict the dealer because the dealer is not allowed to think. The dealer draws until reaching 17 or more and then stops, with the only variation being whether a soft 17 is also drawn to. That fixed procedure means the dealer's upcard determines a known probability distribution over final outcomes: 17, 18, 19, 20, 21, or bust.
Those distributions have been computed exhaustively. The number that matters most for intuition is the bust rate, and it is not smoothly distributed. It has a cliff in it.
| Dealer upcard | Approximate bust rate |
|---|---|
| 2 | 36% |
| 3 | 38% |
| 4 | 40% |
| 5 | 42% |
| 6 | 44% |
| 7 | 26% |
| 8 | 24% |
| 9 | 23% |
| 10 | 23% |
| Ace | lowest of all |
Those are the commonly published figures for a six-deck game in which the dealer stands on soft 17, rounded. They shift by a few tenths of a percent with the deck count and the soft 17 rule. The ace is left unquantified above on purpose: published tables report it two different ways depending on whether hands where the dealer already holds a blackjack are counted, so a single number invites confusion. What is not in doubt is that the ace is the strongest upcard the dealer can show.
The cliff at 7 is the whole chart in one number
Look at what happens between the 6 and the 7. The bust rate falls from roughly 44 percent to roughly 26 percent in a single step, close to cutting in half. Nothing else in the table moves like that.
The reason is mechanical. A dealer showing 6 needs to build a total of 17 or better out of a starting point of 6, which usually takes at least two more cards, and each additional card is another chance to go over. A dealer showing 7 gets to 17 with a single ten-value card, and roughly four of every thirteen cards are ten-valued. The 7 is the first upcard that reaches a safe total in one draw. That is the entire explanation for why every basic strategy chart has a visible seam between the 6 column and the 7 column.
Once you see the seam, most of the chart's shape follows:
- Against 2 through 6 (weak upcards): the dealer is likely to bust, so your job is mostly to not bust first. You stand on stiff totals you would otherwise hit, you double more, and you split more, because busting the dealer does the work for you.
- Against 7 through ace (strong upcards): the dealer will probably make a standing total, so a mediocre hand of yours is likely to lose without improvement. You take more cards, you double less, and you accept more bust risk because standing pat loses anyway.
21 TrainerFree on the App Store. Drill every hand against every upcard, with the reason behind each answer.
Get the app
Weak does not mean the same weak
Grouping 2 through 6 together is a useful first approximation and a poor final one. The 2 busts around 36 percent of the time and the 6 around 44 percent, and eight percentage points is a large gap in a game decided by fractions of a percent. This is why the chart's left edge is not a clean block.
The classic example is 12 against a 2 or a 3. Against a 4, 5 or 6 you stand on 12, letting the dealer take the bust risk. Against a 2 or a 3 you hit, because those upcards do not bust often enough to justify sitting on a total that only wins when the dealer busts. It is one of the closest calls on the chart and one of the most commonly misplayed hands. Doubling boundaries move in the same direction: 9 doubles against some weak upcards and not others, and soft doubles start and stop at different columns.
Treat 5 and 6 as the genuinely weak cards, 4 as clearly weak, and 2 and 3 as weak but not weak enough to be generous with. Our guide to hitting and standing covers where each boundary actually falls.
The 10 is the card you will see most
Ten-value cards are four of the thirteen ranks, so a 10 upcard appears roughly three times as often as any single other rank. This has a practical consequence: your most frequent difficult decisions will all be against a 10. Hard 16 against a 10 is the most discussed hand in blackjack for exactly this reason, and we gave it its own article.
Frequency also means it is where sloppiness costs the most. If you are going to be rigorous about one column of the chart, make it the 10.
Stop guessing the hole card
A persistent piece of table folklore says to assume the dealer's hidden card is a 10. It sounds conservative and it is simply inaccurate. Around nine of every thirteen cards are not ten-valued, so the more likely case is that the hole card is something else. Reasoning from a fixed assumption throws away the information in the actual distribution.
Basic strategy already integrates over every possible hole card, weighted by its real probability, and then over every possible draw sequence after that. That is what makes it optimal. A rule of thumb built on one guess cannot beat a calculation built on all of them.
Where the upcard reading changes
Two things move these numbers. The first is the soft 17 rule: when the dealer hits soft 17, the dealer busts slightly more often but also improves some hands that would have stood on 17, and the net effect favors the house by roughly a fifth of a percent. The S17 versus H17 comparison spells out which hands change.
The second is the count, for players who have gone on to card counting. Bust rates by upcard are averages over a neutral deck. When the remaining deck is rich in ten-value cards, weak upcards bust more and the correct play on several borderline hands flips. That is the entire basis of index deviations, and it is a layer to add only after the base chart is automatic.
For everyone else, the practical instruction is short. Look at the dealer's card before you look at your own, sort it into weak or strong, and let the chart handle the boundaries you have not yet internalized. Then drill the boundaries until you have.
Drill the hard columnsHeat maps show which dealer upcards you keep getting wrong. No account, no ads, works offline.
Download
Frequently asked questions
Which dealer upcard is the weakest in blackjack?
The 6, by a small margin over the 5. Per the commonly published dealer outcome tables for a six-deck game where the dealer stands on soft 17, a 6 upcard busts roughly 44 percent of the time and a 5 roughly 42 percent. Those are the two cards basic strategy responds to most aggressively, with more doubling and more splitting than against anything else.
Why does basic strategy change at a dealer 7?
Because the dealer's bust rate falls off a cliff there. Against a 6 the dealer busts around 44 percent of the time; against a 7 that drops to roughly 26 percent, because a 7 reaches a standing total with one ten-value card instead of needing two cards. That single step is why the chart treats 2 through 6 as weak upcards and 7 through ace as strong ones.
Does the dealer bust more often against a 5 or a 6?
A 6, slightly. The published multi-deck tables put the 6 near 44 percent and the 5 near 42 percent when the dealer stands on soft 17. The practical difference is small, and both are treated the same way by most of the strategy chart. A few specific hands, such as splitting 10s in count-based play, distinguish between them.
Is a dealer ace or a dealer 10 worse for the player?
An ace. It has the lowest bust rate of any upcard and it carries the largest chance of an immediate dealer blackjack, so it threatens the player in two ways at once. A 10 is the second worst and appears far more often, since four of the thirteen ranks are ten-valued, which is why hands like 16 against a 10 come up so frequently.
Should I assume the dealer's hole card is a 10?
No. It is a common shortcut and it is wrong often enough to cost you. Roughly four in thirteen cards are ten-valued, so the hole card is more likely not to be a ten than to be one. Basic strategy already accounts for every possible hole card weighted by its actual probability, which is why following the chart beats reasoning from a single assumption.