When the dealer shows an Ace, the dealer will offer "Insurance" to all players at the table. The bet pays 2:1 if the dealer has blackjack. It costs half your original wager. The casino frames this as protecting yourself against the dealer having a natural — you buy protection, the dealer has blackjack, and you break even on the hand.
It sounds rational. It is not. Insurance is one of the worst bets available at any blackjack table, and understanding why requires a simple calculation that takes about thirty seconds.
Insurance pays 2:1 if the dealer has blackjack — meaning their face-down card is a ten-value card (10, J, Q, or K). In a standard six-deck shoe, there are 96 ten-value cards out of 312 total cards. After the dealer's Ace is removed from the count, there are 311 cards remaining, 96 of which are ten-value cards.
Every dollar wagered on insurance returns an expected loss of 7.4 cents. This makes insurance nearly as bad as American roulette (5.26%) and dramatically worse than the blackjack main game with basic strategy (0.5%). You are spending an extra 7.4% to play a side bet every time you take insurance.
"Even money" is what happens when you take insurance while holding a blackjack yourself. The dealer asks if you want even money — guaranteed payment of 1:1 instead of waiting to see if the dealer also has blackjack, which would result in a push.
Even money is mathematically identical to insurance. You're still betting half your wager at 7.39% house edge. The difference is that it's framed as a certainty versus a gamble, which makes it psychologically appealing. The guaranteed win feels better than risking a push. But over thousands of hands, refusing even money and taking the 1.5:1 blackjack payout when the dealer doesn't have blackjack produces significantly more income than taking even money every time.
Basic strategy is unambiguous on this: never take insurance, never take even money. No card counter takes insurance unless their count specifically indicates the remaining deck is rich enough in ten-value cards to change the math — which requires a very high true count that basic strategy players cannot track.
Insurance exists because it is profitable for the casino. It generates an additional 7.39% edge on a half-wager from a subset of hands. The framing as "protection" is intentional — it sounds responsible and prudent to insure a large bet. In reality it is the opposite of prudent. You are paying extra to participate in a bet where you lose 69% of the time.
Dealers are trained to offer insurance clearly and helpfully. Some will even suggest it if you have a large bet or a strong hand. This is not malice — they're following protocol — but the protocol exists because insurance is good for the casino, not the player.
Card counters who track the true count of ten-value cards versus non-ten cards can reach a count where the remaining deck contains enough tens to make insurance a positive expected-value bet. This requires a true count of approximately +3 or higher depending on the counting system. At that count, there are enough tens remaining that the 2:1 payout becomes favorable.
If you are not counting cards, this exception does not apply to you. Basic strategy players should treat insurance as a simple rule: never take it, under any circumstances, regardless of the size of your hand, the size of your bet, or what the dealer suggests.
Taking insurance in blackjack is paying an extra 7.39% on a side bet at a game you're already playing at 0.5% house edge. It turns the best game on the floor into something approaching a mediocre one. Over a lifetime of blackjack play, refusing insurance will save you a meaningful amount of money. The math is clear and it doesn't change based on your hand, the dealer's visible card, or how you feel about the particular shoe.
Decline insurance every time. Decline even money every time. Let the basic strategy chart make the decisions.
Use our full odds table to compare house edges across every casino game and bet type before you play.