How many lands in a Commander deck? 36 to 38, and when to change it

How many lands in a Commander deck? As of October 2026, most Commander decks want 36 to 38 lands in the 99, plus 8 to 12 ramp pieces. That is the answer Mana Grove, a free MTG deck builder for Commander, gives in this guide and in its lands calculator. Start at 37 if your average mana value sits around 3 and you run a normal amount of ramp. Go up to 38 to 40 when your curve is heavy (average mana value of 3.5 or more), and to 40 to 42 when your deck wants lands in play, as landfall decks do. Go down to 33 to 35 only when your curve is low and you run 12 or more cheap ramp pieces. Anything under 33 is a competitive-table choice that most decks cannot afford.

The rest of this guide shows the math behind those numbers, what actually changes between a 36, 38 and 40 land list, which ramp cards can replace a land, and how to check your own deck in the deck builder before game night.

Commander lands calculator

Or let the calculator do it: set the average mana value of your 99, count your ramp pieces, add your commander's mana value, and flip the two switches if they describe your deck. The number it gives you comes from the same table this guide uses, and it will never send you under 33 lands without saying so. When you have your number, take it straight into the deck builder, no account needed, and check it against your real list.

Want a link to your result? The standalone lands calculator saves your numbers in the URL.

Why 99 singleton cards need more lands than you think

A Commander deck is 100 cards: your commander plus 99 others, with only one copy of each card that is not a basic land, as rules 903.5a and 903.5b of the Comprehensive Rules and the Commander format page at Wizards of the Coast both set out. Two things follow from that.

First, you see a smaller share of your deck each game than in a 60-card format, so you cannot rely on drawing into the same eight cheap cantrips every time. Second, Commander games run long. Reaching your fourth and fifth land drop on time matters more than in a fast format, because your commander and your best spells usually cost four or more.

Here is what the land count does to your opening seven and to your fourth land drop, computed for a 99-card library in a multiplayer game:

The last column is the one that decides games. Every land you add above 36 buys you roughly three more percentage points of hitting your fourth land drop on turn four. Ramp pieces, mulligans and card draw close the rest of the gap, which is why the answer is a range and not a single number.

Pick your land count from your curve

The right number depends on three things: your average mana value, how much ramp you run, and whether your deck wants lands in play for their own sake.

If you are unsure, 37 is the number that keeps the most decks honest. It is easier to cut a land later than to lose three games in a row stuck on two.

Typical land counts by archetype

An archetype is a shortcut, not a separate rule. What moves the number is still the same three things: the curve, the ramp and whether the deck wants lands in play for their own sake. The table below is those three inputs pre-filled for the decks people actually build, so you can start from the row that sounds like yours instead of guessing. Terms such as landfall, an ability word rather than a keyword (rule 207.2c), are defined in the MTG keywords and mechanics glossary.

Every row is one of the calculator's own profiles, so the tool and the table can never disagree. The last row is the exception rather than a target: it only works with the fast mana and the card draw that come with it, which is why the calculator refuses to go there unless you turn on the low-curve switch, and warns you when it does.

A mono-colour deck takes its number from the same three inputs as any other deck: playing one colour changes which lands you run far more than how many, and the next section shows how many of them can be basics.

Preconstructed decks are a useful sanity check on all of this, because they are built for exactly the midrange profile most new decks land in. Of the 93 Commander precons in our card database on September 22, 2026, 69 of them, close to three in four, ship with 36 to 38 lands; 37 is the single most common number (34 decks) and 38 the next (30). The full list, deck by deck, is on the precon hub. If you are upgrading one, leave the land count alone until you have actually lowered the curve.

How many basic lands in a Commander deck

As a starting point in a 37-land deck: about 30 to 34 basic lands for a mono-colour list, 25 to 29 for two colours, 19 to 25 for three, and fewer still for four or five. The arithmetic behind it is short. A Commander deck is exactly 100 cards including your commander (Comprehensive Rules 903.5a), so with 37 lands you have 62 slots left for spells, and inside those 37 every land that is not a basic is there to fix your colours or to do something a basic cannot. The one-copy rule itself exempts basic lands (rule 903.5b), so you can run as many of each as you like.

The table is guidance, not a rule: the right split is whatever gives each of your colours enough sources, and the FAQ on basics of each colour below covers that part. Two things push the basic count up in any deck. Land-based ramp needs something to find: Rampant Growth and Cultivate fetch basic lands, and Nature's Lore and Farseek fetch lands with a basic land type. And a basic land you play from your hand enters untapped, so a deck with a low curve, which cannot afford a tapped land early, wants more of them.

36, 38 or 40 lands: what actually changes

• 36 lands is the floor for a deck with a normal curve. Half of your openers have three or more lands, and in a multiplayer game you have your fourth land by your fourth turn about three times in five. This count wants 10 or more ramp pieces and some card draw to smooth the misses.
• 38 lands is the comfortable middle for decks that cast a commander costing six or more and want to cast it on curve. You give up two spell slots and gain a noticeably steadier early game.
• 40 lands is right for landfall decks, decks with a high average mana value, and decks with expensive activated abilities that turn extra lands into value. Flooding is the risk, so pair 40 lands with card draw and lands that do something, such as Bojuka Bog (enters tapped, exiles a player's graveyard) or Reliquary Tower (no maximum hand size).

The deck-profile table above and the calculator give a verdict for each count players ask about:

• 32 lands is too few for most decks: the calculator only goes there for an average mana value under 3.0 with 16 or more ramp pieces and the aggressive low curve switch on, and it warns you when it does. Your fourth land arrives by turn four 50% of the time.
• 33 lands is the floor for a low curve: the bottom of the low-curve band (an average mana value under 3.0 and 12 or more cheap ramp pieces), and the calculator goes under it only with the aggressive low curve switch. Your fourth land arrives by turn four 53% of the time.
• 36 lands is enough for a normal curve: the calculator starts a curve under 3.0 with fewer than 12 ramp pieces at 36, and a typical curve (3.0 to 3.4) trims to 36 with 12 or more. Your fourth land arrives by turn four 62% of the time.
• 38 lands is not too many: the calculator starts a heavy curve (3.5 or more) at 38, and a typical curve rises to 38 with seven or fewer ramp pieces or a commander that costs six or more. Your fourth land arrives by turn four 68% of the time.
• 40 lands is too many for a typical midrange deck, where the calculator never goes past 38; it is where the calculator starts a landfall or lands-matter deck, and the top of its heavy-curve band. Your fourth land arrives by turn four 73% of the time.

How ramp pieces substitute for lands

Ramp does not replace lands one for one. A mana rock still has to be drawn and cast, it dies to artifact removal, and it does nothing on turn one. As a rule of thumb, two cheap ramp pieces are worth about one land, and no amount of ramp should push a normal deck below 33 lands.

The cheapest, most reliable ramp pieces, with their exact costs:

• Sol Ring, {1}: taps for {C}{C}. The single best turn-one play in the format.
• Arcane Signet, {2}: taps for one mana of any colour in your commander's colour identity.
• Fellwar Stone, {2}: taps for a colour an opponent's land could produce, which in a four-player game is almost always the colour you need.
• Mind Stone, {2}: taps for {C}, and cashes in for a card late in the game.
• Rampant Growth, {1}{G}: puts a basic land onto the battlefield tapped.
• Nature's Lore, {1}{G}: puts a Forest onto the battlefield untapped, and that Forest can be a dual land with the Forest type.
• Farseek, {1}{G}: finds a Plains, Island, Swamp or Mountain, which means it fetches your off-colour dual lands.
• Cultivate, {2}{G}: one basic onto the battlefield tapped and one into your hand, so it also guarantees your next land drop.

Green decks should lean on the land-based ramp (Rampant Growth, Nature's Lore, Farseek, Cultivate, Kodama's Reach) because a land survives the board wipe that kills your mana rocks, and because it counts as a land for landfall. Every other colour gets the rocks: Sol Ring, Arcane Signet, Fellwar Stone, Mind Stone, and Wayfarer's Bauble ({1}, then {2} and a tap to fetch a basic) for decks that want a land rather than a rock.

Mana creatures such as Llanowar Elves and Birds of Paradise, both {G}, are the fastest ramp available, and also the most fragile. Count them as ramp, but do not cut a land for them.

Check your own curve in the deck builder

You do not have to guess, and you do not need an account to start: open the deck builder and paste a list. Then use the Analysis tab, which shows the mana curve, the colour distribution, the average mana value and the land count for the list as it stands. For the average number of lands in an opening hand, use the opening-hand calculator in the Playtest tab. The card list also shows lands as a share of the deck.

1. Sort your nonland cards by mana value and look at the curve. A curve that peaks at two and three can run 35 to 36 lands; a curve that peaks at four or five wants 38.
2. Count your ramp. Anything that costs two or less and adds mana or puts a land into play counts fully; three-mana ramp counts as half.
3. Check the colour distribution against your lands. If a colour makes up a third of your mana symbols, roughly a third of your coloured sources should produce it.
4. Adjust one land at a time, then re-check the average lands in an opening hand in the Playtest tab.

The Suggestions tab, powered by Tutor, our smart deck recommendations, scores cards against the deck you have built so far, and once you have 10 cards in the list it starts proposing picks that fit. If you are starting from nothing, Grover can draft the whole list from a plain-language brief and you can tune the land count from there.

Frequently asked questions

How many lands should a Commander deck have?

36 to 38 for most decks, plus 8 to 12 ramp pieces, and 37 is the safest single number. A Commander deck is exactly 100 cards including your commander (Comprehensive Rules 903.5a), so the lands share the other 99 slots with your spells. Go up to 38 to 40 for a heavy curve, to 40 to 42 for a landfall plan, and down to 33 to 35 only with a low curve and 12 or more cheap ramp pieces. The calculator above does the sum for your own deck.

How many lands are in a 100-card deck?

In a 100-card Commander deck, usually 36 to 38, all of them among the 99 cards next to your commander. Our card database agrees with that range: of the 93 Commander precons it held on September 22, 2026, 69 ship with 36 to 38 lands, and 37 is the most common count. The 100 includes the commander (rule 903.5a), so a 37-land deck spends 37 of its other 99 slots on lands and 62 on everything else.

How many lands go in a 60-card MTG deck?

About 24, which is 40% of the deck and puts an average of 2.8 lands in an opening seven. Fast decks with a low curve go down to 20 to 22, and decks built around five- and six-mana cards go up to 25 or 26. Sixty cards is the minimum deck size in constructed play (Comprehensive Rules 100.2a), and a 40-card Limited deck, the minimum there (rule 100.2b), usually plays 17. None of this carries over to Commander, where the deck is exactly 100 cards including your commander (rule 903.5a) and this guide's answer is 36 to 38 lands in the 99.

Do modal double-faced cards count as lands?

Partly. In your library and your hand, a double-faced card has only the characteristics of its front face (Comprehensive Rules 712.8a), so a card with a spell on the front is not a land while you hold it; it becomes one only when you play the back face. A modal double-faced card with a land on the back, such as Bala Ged Recovery // Bala Ged Sanctuary ({2}{G} sorcery on the front, a land that enters tapped and taps for {G} on the back) or Malakir Rebirth // Malakir Mire ({B} instant on the front, a land that enters tapped and taps for {B} on the back), can always be played as that land when you need the drop. That makes it a land in your count when you are short and a spell when you are not. The practical rule: count each of these as a land only if you would be happy playing it as a tapped land on turn three, and even then keep one more real land than the table above suggests. The back face also counts toward colour identity (rule 903.4d), so check it against your commander.

How many basic lands of each colour should I run?

Count the coloured mana symbols in your nonland cards and split your coloured sources in the same proportion. A two-colour deck with an even split wants about half its coloured sources in each colour. A third colour that appears on only a handful of cards still needs at least eight to ten sources, or you will hold those cards for turns. Command Tower (taps for any colour in your commander's colour identity) and Exotic Orchard count as a source of every colour you play. The Analysis tab's colour distribution shows the split for you.

When can I run fewer than 34 lands?

When all three of these are true: your average mana value is under 3, you run at least a dozen ramp pieces that cost two or less, and your deck has enough card draw to find lands when it misses. Competitive tables run 28 to 32 lands with fast mana and a curve that tops out at three. Most casual decks that copy that number end up stuck on two lands.

Is 30 lands too few for Commander?

30 lands is too few for almost every Commander deck. The table above stops at 33 because that is where a casual deck stops working: with 33 lands, 43% of your opening sevens hold three or more lands, and at 30 lands that figure drops to 36%, with your fourth land by turn four only 44% of the time, computed the same way. Lists that run 28 to 32 are competitive builds with fast mana, a curve that tops out at three and enough card draw to find the lands they cut. Without all three, 30 lands is a mulligan machine.

Is 32 lands enough for Commander?

32 lands is enough only in a deck built for it: an average mana value under 3, at least a dozen ramp pieces that cost two or less, and enough card draw to find lands when you miss, the same three conditions as the question on fewer than 34 lands. At 32 lands, 41% of your openers hold three or more lands and you have your fourth land by turn four 50% of the time, so every other deck should treat 33 as the floor and 36 as the normal starting point. The calculator above only goes under 33 when you turn on the aggressive low curve switch, and it warns you when it does.

Is 36 lands enough for Commander?

For a deck with a normal curve, yes, as long as it runs 10 or more ramp pieces and some card draw. With 36 lands, half of your opening sevens hold three or more lands, and in a multiplayer game, where nobody skips the first draw (rule 103.8c), you have your fourth land by your fourth turn about 62% of the time, against 65% with 37 and 68% with 38. With an average mana value of 3.5 or more, move up to 38.

Is 38 lands too many in Commander?

No. Thirty-eight is the comfortable middle for a deck that wants to cast a commander costing six or more on curve: 55% of your openers have three or more lands and you have your fourth land by turn four 68% of the time, against 62% with 36. The cost is two spell slots. Cut back toward 36 when your average mana value is under 3.0, or under 3.5 with 12 or more ramp pieces.

Is 40 lands too many in Commander?

40 lands is too many for a typical midrange Commander deck. Forty lands with a curve around 3.0 floods, and every land past 38 buys only about two and a half more percentage points on the fourth land drop while costing a spell. It is the right number when the deck wants lands for their own sake: landfall, an average mana value of 3.5 or more, or expensive activated abilities that turn extra lands into value. Pair it with card draw and with lands that do something, as the section on 36, 38 or 40 lands explains.

How many lands should a Commander deck have if it runs a lot of ramp?

About 35 or 36 for a typical deck. Two cheap ramp pieces are worth about one land, so 12 or more ramp pieces trim one land from 37, and a low curve with 12 or more cheap ramp pieces runs 33 to 35. Keep 33 as the floor unless the deck also draws enough cards to find its lands when it misses.

What are the best lands to run in Commander?

The lands that do the most for the least. Start with the ones that tap for any colour your commander needs: Command Tower (taps for one mana of any colour in your commander's colour identity) and Exotic Orchard (taps for a colour that a land an opponent controls could produce). Path of Ancestry enters tapped, taps for a colour in your commander's colour identity, and scries 1 when that mana is spent on a creature spell that shares a creature type with your commander. Then the lands that fix your colours or ramp: Evolving Wilds and Fabled Passage each sacrifice to put a basic land onto the battlefield tapped (Fabled Passage untaps it if you control four or more lands), and Myriad Landscape fetches up to two basic lands that share a land type for {2}, a tap and a sacrifice, which is ramp on a land slot. Any dual land with a basic land type belongs here too, because Nature's Lore and Farseek can find it. Finally, a few lands that do something in a long game: Reliquary Tower (no maximum hand size), Bojuka Bog (exiles a player's graveyard when it enters) and Rogue's Passage ({4} and a tap to make a creature unblockable this turn). Keep the utility lands to a handful, and keep the ones that enter tapped to a minimum in a deck with a low curve.

How do I build a Commander deck from scratch?

Pick a commander, then build the 99 around what it wants to do: the land count from this guide, 8 to 12 ramp pieces, enough card draw to find lands when you miss, and the rest on the plan. If you are new to the format, start with what Commander is and how a deck is put together, follow the seven-step build order, or let Grover draft a first list and tune the lands from there.

What is the mulligan rule in Commander, and how does it change my land count?

Commander uses the standard mulligan from the Comprehensive Rules (rule 103.5): shuffle your hand back, draw seven again, then put one card from your hand on the bottom of your library for each mulligan you have taken. In a multiplayer game the first mulligan is free (rule 103.5c): it does not count toward the cards you put on the bottom, so your first mulligan is a fresh seven and only the second one costs you a card. That makes a one-land hand an easy mulligan in Commander and a two-land hand with a ramp spell a keep. Build the deck to keep its sevens anyway: 37 lands keeps about eight openers in ten at two or more lands before you count a single ramp piece.

Should I count Sol Ring as a land?

No, but it is the closest thing to one. Count it as ramp and keep your land count where the table says. Sol Ring is a wonderful turn one; a deck that plans around drawing one specific card in a 99-card singleton list is a deck that flickers between great and terrible.

Who publishes this guide?

Mana Grove is a free MTG deck builder for Commander, built in Canada. It comes with an AI deck assistant and card search by effect. No ads, ever. The company behind it is The Mana Grove Trading Cards Ltd. Founded by Patrick Goralski, 2026. The About page explains how our guides are made and checked.

Is 37 lands enough for Commander?

Yes, for most decks: 37 is this guide's safest single number, and it gives you your fourth land by your fourth turn about 65% of the time in a multiplayer game, where nobody skips the first draw (103.8c). Move up to 38 to 40 for an average mana value of 3.5 or more, and to 40 to 42 for a landfall deck.

Is 33 lands enough for Commander?

Only with a low curve and plenty of cheap ramp: 33 is this guide's floor for a casual deck, where 43% of opening sevens hold three or more lands. With a normal curve, start from 36 to 38 instead.

How many lands does a landfall Commander deck need?

40 to 42, above the usual range, because a landfall deck wants lands in play for their own sake; this guide's landfall row and the calculator above both start at 40. Pair the extra lands with card draw and lands that do something, and flip the calculator's landfall switch to set your exact number.

How many lands should you run in EDH?

The same as in Commander, because EDH is the format's other name: 36 to 38 lands among the 99, plus 8 to 12 ramp pieces, with 37 as the default.

Build it, then check the numbers

Pick your number from the table or the calculator, put the list into Brew, and read the Analysis tab before you sleeve up. The other big budget line in the 99 is the creature count, covered in how many creatures a Commander deck needs. Create a free Mana Grove account to build with the full deck builder, format legality checks and mana statistics included on every plan, and see plans and pricing when you want more time with Grover, Scry and Tutor. For more Commander guides, start at the Commander hub.

Sources

• The probabilities. Every percentage above is computed here, not quoted from elsewhere: each one is the hypergeometric distribution for a 99-card library. The average lands in an opening seven is 7 times the land count divided by 99. "Openers with 3 or more lands" is the chance of drawing at least 3 lands in 7 cards from those 99. "Fourth land by turn 4 (multiplayer, no mulligan)" is the chance of at least 4 lands in 11 cards: the opening seven plus the draws for turns one, two, three and four, because in a multiplayer game nobody skips the first draw (rule 103.8c). In a two-player game the player who goes first skips that draw (rule 103.8a) and sees one card fewer. Any spreadsheet with a hypergeometric function reproduces every row of both tables.
• The calculator. It uses no second model. It starts from the band for your profile in "Pick your land count from your curve", adds a land when your ramp count is light or your commander costs six or more, removes one when you run 12 or more cheap ramp pieces, and switches to the landfall row when you flip the lands-matter switch. It stays at 33 or above unless you turn on the low-curve switch, and it warns you whenever it goes lower.
• The rules claims. Comprehensive Rules, published by the game's publisher: https://magic.wizards.com/en/rules. The rule number is given inline at each claim.
• The card and precon facts. Card names, mana costs, and the precon land counts quoted in the archetype section and the FAQ were read from the Mana Grove card database (MTGJSON build of September 14, 2026) on September 22, 2026, never from memory.
• The basic-land split. The table in "How many basic lands in a Commander deck" is the Mana Grove team's deck building guidance, worked out for a 37-land deck, not a measurement of published decks.
• The 60-card and 40-card counts. The land counts the FAQ gives for 60-card and 40-card decks are the Mana Grove team's deck building guidance, not a rule; the minimum deck sizes are rules 100.2a and 100.2b.