The finding this page is built on
Three formats settled on three land counts, independently, over decades of play. Nobody coordinated it. And all three deliver almost exactly the same thing:
| Convention | Deck | Chance of making the third land drop |
|---|---|---|
| 24 lands | 60 cards | 78.9% on the play, 85.6% on the draw |
| 37 lands | 99 cards, Commander | 80.0% on the draw |
| 17 lands | 40 cards, limited | 84.5% on the play |
Forty cards, sixty cards, ninety nine cards. Three different games, really. And every one of them arrived at hit your third land drop about four times in five.
That is what makes this page able to compute rather than quote. The target is not something we invented and it is not a coefficient copied out of somebody else's article. It is the number the game already agreed on, recovered from the counts everyone plays, and then applied to your curve instead of an average one.
Why you cannot hit your late land drops
This is the part that surprises people, and it is worth seeing in a table. Here is what a standard 24 land deck actually does, on the play:
| Land drop | On time |
|---|---|
| Turn 2 | 91.0% |
| Turn 3 | 78.9% |
| Turn 4 | 63.2% |
| Turn 5 | 46.7% |
| Turn 6 | 31.9% |
Your sixth land arrives on turn six less than a third of the time. And you cannot buy your way out of it: getting that to 90% takes 38 lands, which is 63% of the deck. At that point you have a deck made of lands that occasionally draws a spell.
So a six mana card is not a card you cast on turn six. It is a card you cast on turn eight, and the honest way to plan a deck is to know that going in. This is the entire reason ramp exists.
Screw and flood pull against each other
Mana screw is too few lands. Flood is too many. Adding lands fixes one and causes the other, which is why the question has an answer at all rather than just going up forever.
| Lands in 60 | Opening hand with 0 or 1 | Opening hand with 5 or more |
|---|---|---|
| 22 | 19.0% | 5.6% |
| 24 | 14.3% | 8.3% |
| 25 | 12.2% | 9.9% |
| 26 | 10.4% | 11.8% |
| 28 | 7.4% | 16.1% |
They cross between 25 and 26. Below that you are more likely to be short than flooded, above it the reverse.
The thing that actually helps, and it is not lands
Here is the number that should change how you think about this. A keepable opening hand means two to four lands. Going from 22 lands to 26 moves that from 75.4% to 77.8%, which is 2.4 points for four whole card slots. You are trading screw for flood almost one for one.
One mulligan, at 24 lands, moves it from 77.5% to about 94.9%.
The mulligan is worth more than any land you could add, and it costs you nothing to build. The same is true of cheap card draw, which is why the published formulas subtract lands for cantrips. If your deck feels like it floods and screws in equal measure, the answer is usually not a different land count. It is more ways to dig.
How this handles ramp and rocks
A mana rock is not a land, and pretending it is one will get you screwed. You have to cast it first, which means spending a turn and needing lands to do it. But after that it makes mana, so it is not nothing either.
We count each ramp spell or rock as half a land and each cheap cantrip as a third. Those are our editorial figures, not published ones, and the page shows you exactly what it subtracted so you can disagree with the arithmetic rather than just with the answer. They sit in the range the common formulas use, and the honest position is that anyone quoting a precise coefficient here is quoting a fit, not a law.
Commander is a different game
The 37 land convention is sound, and it holds up under the maths better than most conventions do. Two things make Commander behave differently from the numbers alone:
- You are on the draw most of the time. In a four player pod, only one seat plays first, so three quarters of your games include that extra card. It is worth roughly two lands.
- Your commander is always castable. A 60 card deck with a bad draw has nothing to do. A Commander deck always has one spell available, which quietly forgives a lot of awkward mana.
Use 99 for the deck size. The command zone is not part of your library, so your commander is never a card you might draw.
This is half the question
How many lands and which lands are two different problems, and this page only does the first. Once you have a total, the manabase calculator works out the colour split from the coloured pips in your casting costs, so you know how many of these lands need to make red.
If you want the raw probability behind any of this, the hypergeometric calculator is the engine both pages run on, and the opening hand calculator covers mulligans and the odds of drawing a specific card.