Why You Should Never Take Insurance in Blackjack

When the dealer shows an Ace, they will ask if you want insurance. It sounds protective: "do you want to cover yourself in case I have blackjack?" That framing sells millions of dollars of losing bets to tourists every year. Insurance is not protection. It is a separate side bet on whether the dealer's down card is a ten — and it pays worse than the actual odds of that card being a ten. It is the worst bet on the table and you should decline it every single time, including when you have a 20, a pair of Aces, or anything else that feels like it "should" take it.

What insurance actually is

When the dealer's upcard is an Ace, you can pay half your original bet as a side bet that the dealer's hole card is a ten-valued card (10, J, Q, or K). If they do have a ten, insurance pays 2:1. If they do not, you lose the insurance bet and then play out the hand normally.

The bet is completely separate from your main hand. Whether you win or lose the original hand has nothing to do with whether insurance wins. People think of them as linked because they happen at the same time, but mathematically they are independent events.

The math in one paragraph

A standard deck has 16 ten-valued cards out of 52. So in a fresh shoe, the probability that any specific unseen card is a ten is about 30.8% (16/52). For insurance to be a break-even bet, it would need to pay out 16 times for every 36 losses — roughly 2.25 to 1. The casino pays 2 to 1. That 0.25 gap is the house edge on insurance, and it works out to about 7.4% per bet. For comparison, the main game at perfect basic strategy runs about 0.5% house edge. Insurance is fifteen times worse.

"But what if I have a 20?"

A 20 is a strong hand — but that has nothing to do with whether insurance is a good bet. Your 20 wins or loses against the dealer's hand regardless of whether you also bought insurance. Taking insurance does not "lock in" your 20. All it does is layer a losing side bet on top of a winning main bet.

The casinos know this, which is why dealers ritually say "insurance?" when the dealer shows an Ace and the player shows a 20. They know you think the 20 is worth protecting. You do not protect a good hand by making a bad bet next to it.

"Even money" is the same mistake in a disguise

If you have a blackjack and the dealer shows an Ace, the dealer will offer you "even money" — 1:1 on your blackjack, instead of the usual 3:2, taken immediately before the hole card is checked. The dealer sells this like you are locking in a sure thing. It is mathematically identical to taking insurance on your blackjack. It is a bad bet in a nicer-sounding package.

The math: if you decline even money and play it out, you get paid 3:2 (1.5 units) on the 69.2% of the time the dealer does not have blackjack, and you push (0 units) on the 30.8% when they do. Expected value: 1.5 × 0.692 = 1.038 units. Taking even money pays exactly 1.000 units. You are giving up 3.8% of your winnings in exchange for certainty — and certainty is not worth 3.8% per hand.

The one real exception

Insurance becomes a good bet only when you know the remaining shoe has a higher proportion of ten-valued cards than normal. In Hi-Lo card counting terms, that threshold is a true count of +3 or higher. At that point, the probability that the hole card is a ten climbs enough to make the 2:1 payout worth taking.

This is the most famous "Illustrious 18" deviation in basic strategy — the full list and the math behind each is catalogued in Don Schlesinger's Blackjack Attack. It is literally the first deviation most counters learn, because it comes up often and has a large impact. But it requires an accurate true count, which requires card counting, which is a separate skill set. (If you want the whole picture, our guide to Hi-Lo counting walks through it.) If you are not counting, you do not know the true count, so this exception does not apply to you. Decline insurance.

Why the mistake keeps getting made

Three reasons:

1. The language is manipulative. "Insurance" is a loaded word. Real insurance is a rational hedge against catastrophic loss. This is a flat negative-EV side bet. Calling it "insurance" triggers the loss-aversion part of your brain.
2. The dealer asks loudly. You are under a mild social pressure to decide quickly in front of other players. The default move when you do not know is usually the one that sounds safer. "Safer" is not "better."
3. Confirmation bias. When you take insurance and the dealer has blackjack, you remember being right. When you take it and they do not, you shrug it off as bad luck. The times it paid off are salient; the times it cost you are forgotten. Over thousands of hands, it is a steady drain.

What to actually say

When the dealer asks, a small head shake or a flat "no" is enough. You do not need to explain. Most experienced players skip insurance so automatically that they do not even hear the question.

If you want to see this in action, our trainer plays through real hands with real counts, and you can feel the difference a few hundred hands of correct play makes. Declining insurance is boring, and boring is exactly what you want from every individual decision at a blackjack table.

The short version

Unless you are counting cards and the true count is +3 or higher, insurance is a losing bet with a 7.4% house edge. It does not protect your hand. It does not lock in a win. It is a separate, bad bet disguised with a reassuring name. Say no, every time.