Blackjack insurance is not protection for your hand. It is a separate wager that the dealer has a ten-value card hidden under an Ace. The usual offer is up to half of your original blackjack wager, and the usual payout is 2:1. That price breaks even only when more than one third of the possible unseen cards are tens, jacks, queens, or kings.
For a player who is not tracking the composition of the remaining cards, that condition is normally not met. The practical default is therefore simple: play your main hand according to the correct strategy and decline insurance as a separate negative-expectation bet.
That answer does not mean insurance never wins. It wins every time the dealer has blackjack. The issue is price: it does not win often enough at 2:1 under ordinary card composition.
The offer you are actually buying
Suppose you bet $20 on the main hand and the dealer shows an Ace. Before the dealer resolves blackjack, you may be offered insurance of up to $10.
The insurance wager and the main wager are then settled independently:
| Dealer result | $20 main hand | $10 insurance wager |
|---|---|---|
| Dealer has blackjack | Main hand usually loses unless you also have blackjack | Wins $20 profit at 2:1 |
| Dealer does not have blackjack | Main hand continues normally | Loses $10 |
The name makes the wager sound like a hedge against losing the $20. Mathematically, however, you have simply added a second $10 bet on one proposition: is the dealer’s hidden card worth ten?
That distinction explains several common confusions. A player with 20 does not have a better insurance bet than a player with 12 merely because 20 is a stronger main hand. A player who has already lost several hands does not receive a better price. A player who “feels” the dealer is due for blackjack does not change the composition of the shoe.
The main hand and the insurance wager happen at the same table, but they answer different questions.
Why 2:1 requires a one-third win rate
Let the insurance stake be one unit. If insurance wins, the player earns two units of profit. If it loses, the player loses one unit.
At break-even, expected profit is zero:
[ 2p - (1-p) = 0 ]
where:
- (p) is the probability that the dealer’s hidden card is a ten-value card;
- (2p) is the expected profit contribution from wins at 2:1;
- ((1-p)) is the expected loss contribution from losing insurance bets.
Solving:
[ 2p - 1 + p = 0 ]
[ 3p = 1 ]
[
p = \frac{1}{3} = 33.33%
]
So the insurance wager becomes favorable only when the conditional probability of a ten-value hole card is greater than 33.33%.
Another way to see the same threshold is with card counts. If there are (T) unseen ten-value cards and (N) unseen non-ten-value cards, insurance is favorable when:
[ 2T > N ]
That is simply the 2:1 payout translated into composition terms. The number of tens must be more than half the number of all other cards combined.
A fresh shoe normally falls short
A standard 52-card deck contains 16 ten-value cards: four tens, four jacks, four queens, and four kings. There are 36 other cards.
Before accounting for any exposed cards, the ten-value proportion is:
[ \frac{16}{52} = 30.77% ]
That is below the 33.33% break-even point.
In a six-deck shoe, the starting composition is 96 ten-value cards among 312 cards. If the only visible card were the dealer’s Ace, 311 unseen cards would remain, including all 96 ten-value cards:
[ P(\text{ten-value hole card}) = \frac{96}{311} \approx 30.87% ]
For a one-unit insurance wager, expected value would be:
[ EV = \left(\frac{96}{311}\times 2\right) - \left(\frac{215}{311}\times 1\right) ]
[ EV = \frac{-23}{311} \approx -7.40% ]
The exact percentage changes with every visible player card and with the number of decks, so 7.40% is an illustrative fresh-shoe calculation, not a universal house edge for every insurance offer. The central point survives those details: ordinary undepleted composition has too few ten-value cards for a 2:1 insurance price to be attractive.
Your own cards are part of the information set
The dealer’s upcard is not the only exposed card. Your cards and the other players’ cards have also been removed from the pool of possible hole cards.
That means an exact finite-deck insurance calculation should use the actual remaining composition, not the shortcut “four ranks out of thirteen.”
Consider a single-deck illustration after the deal. The dealer shows an Ace. Your hand is 8-7, and suppose no other cards are visible. Fifty cards are unknown. All 16 ten-value cards remain unseen, so:
[
P(10\text{-value}) = \frac{16}{49}
]
Why 49 rather than 50? One of those unknown cards is the dealer’s hole card and, conditioning on the known exposed cards, there are 49 cards that could occupy that position after removing the three visible cards from the 52-card deck. The result is about 32.65%, still below break-even.
Now change your hand to K-Q. Two ten-value cards are already visible and therefore cannot be the dealer’s hole card. Fourteen ten-value cards remain among the 49 unknown positions:
[ P(10\text{-value}) = \frac{14}{49} \approx 28.57% ]
The insurance price is worse, not better, even though K-Q is a strong main hand. This is a useful reminder that insurance depends on card composition, not main-hand strength.
Why a $10 insurance win can hide a losing decision
Insurance has a psychological advantage that the math does not share: when it wins, it often wins at the exact moment the main hand loses to dealer blackjack.
With a $20 main wager and $10 insurance bet:
- dealer blackjack causes the $20 main wager to lose;
- the $10 insurance bet wins $20 profit;
- the two cash outcomes can offset each other.
That feels like successful protection. But the same insurance wager also loses every time the dealer does not have blackjack, and those losing offers occur more often under normal composition.
A hedge can reduce the size of a particular bad outcome while still lowering expected value overall. Those are not contradictory statements. Variance and expectation are different quantities.
If your goal is simply to understand the money movement, read Blackjack Payouts. If your goal is to decide whether a particular offer has crossed the mathematical threshold, use When to Take Insurance as the decision-focused companion page.
Even money is insurance expressed as a guaranteed settlement
When you have a natural blackjack and the dealer shows an Ace, some tables offer “even money.” You accept a 1:1 profit immediately instead of waiting to see whether the dealer also has blackjack.
Under a standard 3:2 blackjack payout, this is economically the same insurance problem expressed differently.
Example with a $20 blackjack wager:
- accepting even money guarantees $20 profit;
- declining even money preserves the normal $30 blackjack profit if the dealer does not have blackjack;
- if the dealer also has blackjack, the normal result is a push.
The decision therefore depends on the same hidden-card probability. If dealer blackjack is less likely than one third, giving up the 3:2 upside is too expensive on average. If the ten-value probability genuinely exceeds one third, the insurance/even-money side of the decision can become favorable.
The detailed settlement proof belongs on Why Basic Strategy Usually Rejects Even Money, where the full 3:2 comparison is worked through without turning this insurance page into a duplicate.
Table procedure: the decision comes before the hand continues
Insurance is usually offered only when the dealer’s exposed card is an Ace. The table pauses for the decision before normal player action resumes.
In a common hole-card game, the sequence looks like this:
- Initial cards are dealt.
- Dealer shows an Ace.
- Insurance is offered, normally up to half the original main wager.
- Players accept or decline.
- The dealer checks the hole card according to the approved procedure.
- Insurance is settled.
- If the dealer does not have blackjack, ordinary play continues.
The exact physical procedure varies. Some games use a card-reader device so the dealer can check for blackjack without seeing the rank of a non-blackjack hole card. Some games do not take a hole card until later. Some electronic or streamed games resolve the same decision through software rather than chips on an insurance line.
Those procedural differences matter, but they do not change the basic 2:1 threshold. A different dealing sequence does not make a bad insurance price good.
Pennsylvania’s current blackjack regulation provides a useful concrete example: when the dealer’s first card is an Ace, it permits an insurance wager of up to half the initial blackjack wager and specifies that the wager wins when the hole card is a king, queen, jack, or ten. The same rule set also allows even-money treatment when selected in the casino’s approved rules submission. See the Pennsylvania blackjack insurance rule. Other jurisdictions and game variants can differ, so the posted table rules remain controlling for the game in front of you.
When counting changes the answer
Basic strategy normally assumes you do not know the exact composition of the undealt cards beyond the information visible in the current hand. Card counting changes that information set.
A balanced count such as Hi-Lo tracks whether disproportionately many low cards or high cards have already left the shoe. If enough low cards have been removed, the undealt cards can become rich enough in tens that insurance crosses its one-third break-even point.
This is one reason insurance is unusual among blackjack decisions. The main hand’s basic-strategy action might remain unchanged while the insurance side bet flips from negative to positive because the count says something specific about ten density.
But a vague “positive count” is not enough. The count has to be converted correctly, the insurance index has to match the counting system, and the player has to know how that system treats rounding and true-count conversion. The detailed decision belongs on When to Take Insurance, while True Count Conversion explains how a running count becomes a per-deck signal.
A player who is not already executing a validated counting system should not use the existence of a counting exception as a reason to improvise. Guessing that a shoe “looks rich in tens” is not card counting.
Three mistakes that make insurance look better than it is
“I have 20, so I should protect it”
Twenty is strong against many dealer outcomes, but the insurance wager is settled solely by the dealer’s hidden card. Your 20 can actually make insurance slightly worse if it contains ten-value cards that are now unavailable to be the hole card.
“The dealer has had several non-blackjacks, so one is due”
Past outcomes do not create a balancing obligation. Only the actual remaining composition can change the next-hole-card probability in a finite shoe.
“Insurance paid me last time”
A winning result does not prove a positive expected value. A 30% event happens frequently. The question is whether the payout compensates for the roughly 70% of times the wager loses.
These mistakes all replace a probability question with a story about the main hand, recent results, or emotion.
Insurance can increase the cost of a session quickly
The insurance wager is optional extra action. If the base bet is $50, the maximum insurance wager is commonly $25. Ten insurance offers accepted at the maximum add $250 of side-bet action even though the base blackjack wager never changes.
Expected loss can be written as:
[ \text{Expected Loss} = \text{Insurance Action} \times \text{House Edge on Insurance} ]
If an illustrative set of offers had a 7% house edge and you placed $250 in total insurance action, the expected loss attributable to insurance would be:
[ 250 \times 0.07 = $17.50 ]
That is not a prediction of what the session will lose. Actual results can be much higher or lower. The calculation only shows why repeated optional wagers matter: every extra bet creates another stream of expected cost.
If you want to measure that distinction more broadly, Expected Value and House Edge explain why actual session results and long-run expectation should not be confused.
A compact way to handle the Ace upcard
For a non-counting player, the clean process is:
- identify insurance as a separate wager;
- do not let the strength of your hand decide it;
- decline it under ordinary basic-strategy assumptions;
- then play the main hand according to the correct chart for the table’s rules.
For an experienced counter, replace the third step with a validated composition test or counting-system insurance index. Do not replace it with intuition.
That is all the decision needs. The table’s speed and the emotional discomfort of seeing an Ace do not improve the payout.
The half-bet limit is what makes the hedge look complete
Why is insurance commonly capped at half the main wager? The answer is tied to the 2:1 payoff.
If the main wager is $40, half is $20. When the dealer has blackjack:
- the $40 main wager normally loses;
- a $20 insurance wager wins $40 profit;
- the $40 insurance profit offsets the $40 main-hand loss.
That settlement is why the wager acquired the name “insurance.” The maximum stake is sized so that a winning insurance bet can offset one ordinary main-bet loss to dealer blackjack. But the sizing does not improve the price. It only describes how large the hedge is allowed to be.
If you insure only $10 against the same $40 main bet, a dealer blackjack produces $20 insurance profit while the main hand loses $40. The net result is a $20 loss. Partial insurance is therefore a smaller version of the same side bet, not a different strategy.
Because expected value scales linearly with stake, taking half as much insurance roughly halves both the long-run expected loss and the short-term swings attributable to that wager. It does not change whether the wager itself is favorable.
“2:1 payout” and “200% return” are not the same wording
Casino payout language can be easy to misread. A 2:1 insurance payout normally means two units of profit for every one unit wagered, with the original insurance stake also returned.
On a $10 winning insurance wager:
- original stake returned: $10;
- profit: $20;
- total chips returned from the insurance position: $30.
The break-even calculation uses the $20 profit versus the $10 loss, which is why the required probability is one third. Treating the total $30 returned as “three-to-one profit” would produce the wrong threshold.
This distinction is the same one explained more generally in Blackjack Payouts: the returned stake is not casino profit paid to the player.
Insurance is independent of the normal blackjack payout
The standard insurance wager can still pay 2:1 even at a table where a natural blackjack pays 6:5 rather than 3:2. Those are two separate pricing rules printed on the same game.
This matters when comparing tables. A 6:5 blackjack payout substantially worsens the value of the main game, but it does not automatically change the one-third break-even threshold on a separate 2:1 insurance wager. Conversely, a normal 3:2 main game does not make insurance favorable.
Even money requires more care because it is a way of settling a player blackjack against an Ace upcard. The familiar equivalence between taking even money and insuring a 3:2 blackjack depends on the underlying blackjack payout and the game’s exact rules. If a table uses a nonstandard natural payout, do not import a memorized even-money comparison without checking the posted procedure.
The practical table-selection lesson is broader: read each material rule separately. House Edge by Rules explains why a favorable rule in one part of the game does not cancel an unfavorable rule somewhere else.
No-peek games change exposure, not the insurance price formula
In some blackjack formats the dealer does not take or check a hole card before players act. Dealer blackjack may therefore be discovered only after players have hit, doubled, or split. That can make the main-game exposure very different from a U.S.-style peek game.
The insurance wager still asks a composition question: how likely is the dealer’s eventual second card to be worth ten? If the wager pays 2:1, the mathematical break-even probability remains one third.
What changes is the surrounding main-hand risk. Under a full European no-hole-card loss rule, additional money placed on doubles or splits may also be lost if the dealer later makes blackjack. Under an “original bets only” treatment, added split or double money may be returned. Those are materially different games.
Do not use insurance as a substitute for understanding that rule. Check Blackjack No-Peek Rule before applying a U.S. strategy chart to a no-hole-card table.
A composition example with other players at the table
Visible cards from other hands matter because they are no longer candidates for the dealer’s hidden card.
Suppose a single-deck game has reached the insurance decision. The dealer shows an Ace. Across the player hands you can see these six cards:
- K, Q, 10;
- 6, 5, 3.
Including the dealer Ace, seven cards are exposed. Forty-five cards are unseen. Three ten-value cards have already been removed, so 13 ten-value cards remain.
The conditional probability is therefore:
[ P(10\text{-value}) = \frac{13}{45} \approx 28.89% ]
That is clearly below one third.
Now imagine instead that the six player cards were 2, 3, 4, 5, 6, and 7. No ten-value cards have been removed. Sixteen tens remain among the 45 unseen cards:
[ P(10\text{-value}) = \frac{16}{45} \approx 35.56% ]
In that artificial single-deck snapshot, the insurance wager would be favorable at 2:1 because the observed low cards have pushed the ten density above the break-even point.
This example is useful because it shows what “composition matters” actually means. It is not mystical pattern recognition. It is a changing numerator and denominator.
Why insurance can be a good test of whether a counting system is real
A player who says “I count cards” but cannot state the insurance threshold for the chosen system is missing one of the clearest composition-sensitive decisions in blackjack.
Hi-Lo, for example, gives ten-value cards and Aces the same -1 tag even though insurance cares about tens and not Aces. Despite that simplification, published Hi-Lo systems use a true-count insurance index because the count correlates strongly enough with ten density to make the decision useful.
But systems differ. Some side counts track Aces separately; some counts are unbalanced and do not use the same true-count procedure; some index numbers assume a particular rounding convention. An index copied from one system should not be transplanted into another.
For that reason this page stops at the structure of the wager. Hi-Lo System explains the tags, and True Count Conversion explains the denominator. The specific “take it or leave it” threshold remains on When to Take Insurance.
Responsible play
Insurance can be especially tempting after a large bet because it offers the feeling of making the result safer. That feeling can lead to more money being placed at risk precisely when a player is already anxious about the main wager.
Casino play should be treated as paid entertainment, not income, investment, or debt recovery. If insurance, side bets, or bet increases are becoming a way to chase losses or avoid the emotional discomfort of a possible losing hand, stopping the session is a better risk-control decision than adding another wager.
The National Council on Problem Gambling help-by-state directory lists support resources in the United States. Elsewhere, use the responsible-gambling or public-health service for your jurisdiction.
The 33.33% decision test in one line
At a 2:1 payout, insurance breaks even when the dealer’s hidden card is ten-value more than one third of the relevant unseen cards. If the probability is below 33.33%, the insurance wager has negative expected value; if a valid composition estimate places it above that threshold, the wager can become positive.
That test is independent of whether your main hand is weak or strong. Even money on a player blackjack is the same insurance probability question expressed through a guaranteed settlement. For a non-counter using ordinary composition, the practical default remains to decline. A counting exception requires a system that actually estimates the remaining concentration of ten-value cards; recent dealer results or a “due” feeling do not supply that information.
Author / Editorial Note
This page is written from a land-based casino operations perspective. The goal is not to make blackjack sound easy or unbeatable. The goal is to separate table procedure, emotional pressure, and expected-value math so players can understand what they are actually buying when the dealer says, “Insurance?”
Bottom line
Insurance is a separate 2:1 wager on the dealer’s hidden card being worth ten. The price requires a probability above 33.33% to be favorable, while ordinary undepleted shoe composition is normally below that threshold. For players who are not accurately tracking remaining-card composition, declining insurance is the sound default. A count can change the answer, but a strong hand, a losing streak, or a feeling about the dealer cannot.