What the decision really is
A mulligan lets you shuffle your opening hand back and draw a new one, usually with a penalty. The details differ between games: some take a card away, some let you draw more and then discard. The core question is always the same. Is this hand, with the chance of improvement it offers, better than a fresh hand that is one card smaller or otherwise worse?
Framing it as a comparison removes much of the emotion. A hand does not need to be good in the abstract; it needs to be better than the alternative. A weak seven-card hand can still be right to keep if the replacement is likely to be worse.
Land counts in a 60-card deck
Take a 60-card deck with 24 lands and a seven-card opening hand. Counting the ways of drawing each number of lands gives these approximate chances:
| Lands in hand | Approximate chance |
|---|---|
| 0 | 2.2% |
| 1 | 12.1% |
| 2 | 26.9% |
| 3 | 30.9% |
| 4 | 19.6% |
| 5 | 6.9% |
| 6 or 7 | 1.4% |
Hands with two, three or four lands together come to about 77 percent. That is why those hands are often described as the keepable middle: they are the most common outcomes and they have access to resources without being flooded.
The expected number of lands is 7 × 24 ÷ 60 = 2.8, which fits the table, as the largest groups sit around that value.
Hands at the extremes
Zero or one land is risky. With no lands you cannot play anything, and with one you depend on drawing more. Hands with six or seven lands have the opposite problem: you can cast what you draw but you hold few spells.
But “risky” is not the same as “always mulligan.” With one land and several cheap spells, a player on the draw has an extra card and more chances to find a second land. With two lands and a hand of expensive spells, the keep may be worse than it looks. Counting lands is a first filter, not a rule.
Quantifying the chance of hitting land drops
Suppose you keep a two-land hand on the play. You want a third land by turn three. On the play you do not draw on turn one, so by turn three you have drawn two more cards. The remaining 53 cards contain 22 lands (24 minus the two in hand). The chance that neither of the next two draws is a land is (31 ÷ 53) × (30 ÷ 52), which is roughly 0.338. So the chance of at least one land in two draws is about 66 percent. On the draw, with three cards seen, the chance rises to roughly 80 percent.
These are only approximations, and what matters is the structure: the probability of finding a missing piece grows with the number of draws you see. That is why being on the draw, with an additional card, supports keeping slightly weaker hands. In games where the player going first skips the first draw, the difference is real.
The cost of going down a card
The penalty for redrawing varies. In a simple version, you start with six cards. Fewer cards means fewer options and less chance of having what you need, but a better chance of a functional mix. A six-card hand with a good land balance is worth more than a seven-card hand that cannot cast spells.
A reasonable approach is to ask two questions:
- If I keep, what is the chance this hand functions by the time I need it to?
- If I redraw, what is the chance the new hand is meaningfully better, given I will have one fewer card?
When both answers are close, many players lean toward keeping, because a smaller hand has a cost that is easy to underrate. This is a tendency, not a theorem.
Hand quality beyond land count
Lands are the easiest part to quantify, but not the only one. Consider:
- Does the hand have plays? A hand of three lands and four expensive cards is slow.
- Does it fit the deck’s plan? An aggressive deck wants cheap pressure, so a slow hand may be wrong even if the lands are fine.
- Is it resilient? A hand that depends on one specific card is fragile.
The curve of the deck matters here, since a deck with a low mana curve is usually happy with fewer lands in hand. The same logic applies to limited drafts, where decks run fewer lands: in a 40-card deck with 17 lands, a seven-card hand has no land about 1.3 percent of the time and one land about 9.2 percent of the time.
Human tendencies
Players often make two kinds of mistake. One is keeping a poor hand because redrawing feels like a loss. This is related to loss aversion, the tendency to weigh losses heavily. The other is redrawing a perfectly serviceable hand in the hope of a spectacular one. A fair-to-good hand kept is usually better than a gamble on a better one, though the game format shapes how much.
It also helps to avoid judging a decision by its result. If you keep a reasonable hand and miss your land drops, the decision may still have been right. A single outcome is little information, as discussed in results thinking. Over a large number of games the percentages above are what matter, and even then only on average.
A short checklist
- Count lands first: two to four in seven cards is usually the comfortable range for this deck.
- Check whether you are on the play or the draw.
- Compare your spells to your resources.
- Think about what the opponent’s deck is likely to do.
- Compare against a smaller hand, not a perfect one.
No set of figures makes mulligans risk-free. A fresh hand is random, and sometimes both options are poor.