August 7, 2026 · 6 min read · blackjack · house edge · table selection

Dealer hits soft 17: H17 vs S17 costs you 0.22%

Yes — at most blackjack tables the dealer hits soft 17, and the rule costs you 0.22 percentage points of house edge. It is printed on the felt in small type, it changes nothing you can see while the hand is being dealt, and it is the second most expensive line on a modern blackjack layout.

Zero point two two sounds like a rounding error until you notice what it is a fraction of. A decent six-deck game runs a house edge well under one percent, so this single rule is a real slice of the whole thing. And unlike a payout change, it doesn't announce itself. The dealer turns over 17 and then just keeps going — one more card, on a hand that already got there.

What a soft 17 actually is

A soft hand is one holding at least one ace that can still count as either one or eleven without busting. A soft 17 is, in the Wizard of Odds' phrasing, an ace and any number of cards totaling six points. The canonical version is A-6. So are 6-A, A-2-4, A-3-3 and A-2-2-2. What they share is optionality: the ace is currently doing the work of eleven, and it can quietly demote itself to one whenever that becomes convenient.

A hard 17 — 10-7, 9-8, 5-4-8 — has no such escape hatch, and no dealer anywhere hits it. The soft-17 rule touches exactly one class of hand, which is why it is so easy to underrate. You will see it come up a few times an hour at most. It just wins slightly more often than it loses when it does.

The two possible rules have standard names and standard abbreviations. S17 means the dealer stands on all 17s, soft ones included. H17 means the dealer hits soft 17 and stands on everything else 17 and above, hard or soft — soft 18 is a stand under both rules. Every other dealer rule is identical: draw to 16 or less, stand on hard 17 or more, no choices, no judgement.

Why hitting it is a free re-roll

Here is the mechanism, and it is arithmetic you can check yourself rather than a number I am citing. Take the dealer's A-6 and deal exactly one more card. Every rank in the deck produces one of three outcomes, and none of them is a bust.

That last point is the whole rule. The dealer holding a soft 17 cannot bust on the next card, because the worst possible draw — a ten — simply converts the ace back to one and leaves a hard 17, which is where the dealer already was. Standing on soft 17 is a guaranteed 17. Hitting it is a guaranteed 17 with a lottery ticket stapled to it.

The lottery pays out roughly three times in ten on the first draw, and another three in ten leave the hand no worse than it started. The catch — the reason this isn't a pure gift — is the middle branch. Draw a 5 through 9 and the ace demotes, the dealer is sitting on a hard 12 through 16, and now the drawing rules force more cards and real bust risk. That branch is where H17's downside lives, and it is smaller than the upside.

Ace, 2, 3 or 4 — four of thirteen ranks, about 31% — turns the hand into a soft 18, 19, 20 or 21 immediately.
A ten, jack, queen or king — four of thirteen ranks by count, so also about 31% — makes it a hard 17. The dealer stands, exactly where they started. The draw cost nothing.
A 5, 6, 7, 8 or 9 — five of thirteen ranks, about 38% — drops the ace to one and leaves a hard 12 through 16. Only now can the dealer bust, and only by drawing again.

The numbers, measured over a whole shoe

Mike Aponte published the comparison for a six-deck shoe, showing how the dealer's final hand is distributed across all dealer hands under each rule. The bust rate does go up when the dealer hits soft 17 — from 29.60% to 30.00%. That is the part players notice, and it is why the rule sometimes gets defended at the table as good for us, really.

It isn't, because the busts are the smaller half of the change. Under his figures the dealer's chance of finishing on exactly 17 falls from 15.25% to 14.02%, a drop of about 1.2 points. Of that, roughly 0.4 points goes into extra busts. The other 0.8 goes into stronger hands: 18 rises from 14.62% to 14.82%, 19 from 14.04% to 14.24%, 20 from 18.85% to 19.06%, and 21 from 7.65% to 7.86%.

Two extra good hands for every extra bust. That ratio is the entire argument, and it explains why the casino wrote the rule the way it did. Seventeen is a bad hand for the dealer — it loses to four player totals and ties one. Trading a chunk of those 17s for a mix of 18-through-21 and busts, at two-to-one in favour of the good outcomes, is a straightforwardly profitable swap.

What 0.22 points buys the casino

The Wizard of Odds maintains a table pricing each blackjack rule as a change to the house edge, measured against a stated baseline of eight decks, dealer stands on soft 17, double on any first two cards, double after splitting, split to four hands. The row reads: dealer hits on soft 17, minus 0.22%. Importantly, the page notes the figures already account for proper basic strategy adjustments — so this is the net cost after you play the H17 chart correctly, not before.

Scale matters here, so put it against its neighbours on the same table. Moving from eight decks to single deck is worth plus 0.48% to the player; double deck, plus 0.19%. Being forbidden to double at all costs 1.48%. And paying 6-to-5 on a natural costs 1.39% — roughly six times the soft-17 rule, which is why that payout gets its own post on this site and this one doesn't need to relitigate it.

There is a genuinely useful comparison hiding in those first two figures. Single deck is worth more than H17 costs: plus 0.48 against minus 0.22 nets out to plus 0.26 in the player's favour. On those two lines alone, a single-deck H17 game beats an eight-deck S17 game. (Two caveats, both mine rather than the source's: these increments are calibrated to that eight-deck baseline and are not strictly additive, so treat the sum as a direction rather than a decimal.) The reason that trade is rarely available in practice is that low-deck games are overwhelmingly the ones paying 6-to-5, and 1.39 swamps everything above.

In dollars the rule is modest, which is the honest way to describe it. Using 70 hands an hour — the Wizard's comp-purposes figure, which he pairs there with a 0.75% realized edge that bakes in player mistakes, so I am borrowing the pace and not the edge — a $25 bettor gives up 25 x 70 x 0.0022, about $3.85 an hour, to the soft-17 rule alone. That is close to four dollars an hour you would rather keep, not a catastrophe. Its real significance is that it stacks: H17 almost never arrives alone.

How to spot it, and how often you'll have to

This one is genuinely easy, because the casino tells you. The rule is printed on the layout, and there are only two things it can say: 'Dealer must stand on all 17s' or 'Dealer hits soft 17.' John Grochowski, who has written about casino games for close to four decades, puts it plainly — it will be noted on the blackjack table if the dealer hits soft 17. Read the felt before you buy in and you are done in three seconds.

Assume the answer is H17 unless the table says otherwise. Grochowski describes hit-soft-17 games as having become the standard even with six decks in play. Vegas Advantage's central Las Vegas Strip survey, last updated May 18, 2026, gives the same instruction to its readers: assume that a casino's game hits on soft 17 unless otherwise noted. Their property-by-property notes bear it out — the S17 games they flag tend to sit in high-limit rooms, with Aria's entry noting that the games in High Limit stand on all 17s.

That is the same migration pattern as the 3-to-2 payout. The better rule still exists; it moved up-market, behind a bigger minimum. Which means for most players most of the time, H17 is not a table to avoid — it is the baseline condition you are playing against, and the useful move is to know how it changes your play rather than to hunt for its absence.

The plays that actually change

Six cells. That is the whole difference between the H17 and S17 charts for a four-to-eight-deck game, per the Wizard of Odds' strategy pages. Take the standard chart and apply these modifications:

These pull in two directions, and it is worth seeing why. The doubles are opportunities: doubling 11 against an ace only becomes correct because an H17 dealer can now reach soft 17 and then break it. The surrenders are the opposite — concessions. An H17 dealer showing an ace ends up with a stronger hand overall, so bailing out for half a bet becomes the least-bad option on hands that were marginal before. That is also why the 0.22-point figure is already net of them: without these six plays the rule would cost you more.

Grochowski's list of the soft-hand and surrender changes matches the Wizard's, which is a useful cross-check on the two that people most often get wrong: soft 18 versus a 2, and soft 19 versus a 6. Standing on a 19 feels obviously right. Against an H17 dealer showing a 6, doubling it is the better play, and it is the single most counterintuitive cell on either chart.

If you only intend to memorise one chart, the Wizard is specific about which: 'If you play a mixture of six-deck games, some where the dealer hits a soft 17, and some where he stands, and you only wish to memorize one strategy, I would recommend you memorize the one where the dealer stands on soft 17. The cost in errors due to playing the wrong strategy is 2.3 times higher playing a stand on soft 17 game, with the hit on 17 strategy, than vise versa.' In other words: if you're going to be wrong, be wrong in the S17 direction.

Surrender 15, a pair of 8s, and 17 vs. dealer A (where late surrender is offered).
Double 11 vs. dealer A, instead of hitting.
Double soft 18 (A,7) vs. dealer 2, instead of standing.
Double soft 19 (A,8) vs. dealer 6, instead of standing.

Where it belongs in your checklist

Order of operations, biggest number first. Check the payout — 3-to-2 versus 6-to-5 is worth 1.39 points and takes one glance. Then check soft 17, worth 0.22 and printed in the same paint. Then the doubling and splitting restrictions, which usually are not printed and which the dealer will tell you if you ask. That is a two-minute lap of the pit, and it settles more of your expected result than anything you will do with the cards afterwards.

Then play the right chart for the table you actually chose. The strategy chart on this site is generated from the rules you set, so flipping the 'Dealer hits soft 17' switch in settings re-computes the hard and soft totals and the chart footer states back which rule set it is showing you. Two caveats worth knowing: the pair rows are the same under either rule, and the chart currently applies the hard-total and doubling adjustments rather than the full set above — the soft 18 double and the 8,8 surrender are still on you to remember.

None of this makes blackjack a good bet. It makes it an accurately priced one, which is the only thing worth optimising when the whole edge lives in fractions of a percent. The dealer hitting soft 17 is not a scandal. It is just a rule that most players have never once read, working quietly, a few times an hour, for decades.

Set your table's rules once and the chart follows — decks, soft 17, double-after-split, surrender — so the play you drill is the play that's correct at your game.

Open the rules-aware strategy chart →

Share this post — the link unfurls with its own card.