Commander guide · September 2026
How many lands does a Commander deck need?
Most Commander decks want 36 to 38 lands. To show what each extra land actually buys, we played the same deck 20,000 times at every land count from 32 to 40.
The short answer
- 37 or 38 lands for a typical deck with about ten ramp pieces and a commander costing 4 or less.
- 38 to 40 if your commander costs 5 or more, or if you play little ramp. Past that, fix the ramp instead: lands can't replace it.
- 35 or fewer only with a low curve and plenty of cheap ramp and card draw. Frank Karsten's formula below tells you how far to go.
What one extra land actually buys
We used a well-built Trostani, Selesnya's Voice deck: average mana value 3.5 and 12 ramp pieces (mana rocks, mana dorks and land ramp), with a commander that costs 4. For each land count, basic lands were swapped in or out for spells near the deck's average cost, and every version played the same 20,000 shuffles, so the only difference is the land count.
Going from 32 to 38 lands cut screwed games from 5.7% to 2.6% and raised the chance of casting Trostani on turn 4 from 74% to 78%. Past 38 the commander came down no faster (78% at 40 lands); the extra lands only showed up as flood, which rose to 4.6%.
Land count also changes your mulligans. With the same keep rule, the deck kept its opening seven in 71% of games at 32 lands and 81% at 38. Fewer mulligans is part of what those lands buy.
The flood numbers look small because the goldfish counts a game as flooded only if it's holding nothing but lands on turn 6. In a real game, drawing lands while your opponents develop feels bad well before that point, which is why the sweet spot sits at 37–38 and not at 40.
What changes the answer
We ran the same test on two variations of the deck: one with its 12 ramp pieces swapped for 2- and 3-mana creatures, and one with a 6-mana commander (Tolsimir, Friend to Wolves) leading the same 99 cards.
Ramp matters more than lands. Without ramp, Trostani came down on curve in 53% of games at 36 lands, against 77% with it. Four more lands only brought that back to 57%. If your commander is late, add ramp before you add lands.
An expensive commander wants more lands. With Tolsimir, every extra land kept helping, about 1.4 points of on-curve rate per land, from 84% at 36 lands to 89% at 40, with flood still only 4.0%.
Karsten's formula
Frank Karsten, who has written extensively on the math of Magic mana bases, published this rule of thumb for Commander:
Average mana value counts your nonland cards, not the commander. "Cheap" ramp and card draw are the ones you can cast early; FreeMully counts those costing 3 or less. For the Trostani deck: 31.42 + 3.13 × 3.51 − 0.28 × 13 = 38.8 lands, right where our simulation stopped improving. Each cheap ramp or draw spell is worth about a quarter of a land, so ten of them let you cut roughly three lands.
A few caveats
These are goldfish numbers: the deck plays alone, with no opponents, removal or counterspells. They measure how well your mana and curve work, not whether the deck wins. Your own deck may land somewhere else: landfall decks and decks with mana-hungry abilities often run more, while low-curve decks with lots of card draw can go lower. The only way to know is to test your own list.
Test your own deck. Paste your list into FreeMully and it plays 10,000 games. The Tuning tab's land-count sweep runs this same test on your deck.
Open the simulatorThe full numbers
Show every result (20,000 games per row)
| Lands | Commander on curve | Avg. commander turn | Screwed | Flooded | Kept first 7 |
|---|---|---|---|---|---|
| 32 | 74.3% | T4.17 | 5.7% | 1.1% | 71.3% |
| 33 | 75.0% | T4.14 | 5.1% | 1.4% | 73.4% |
| 34 | 75.9% | T4.11 | 4.5% | 1.7% | 75.1% |
| 35 | 76.5% | T4.10 | 4.0% | 2.0% | 76.6% |
| 36 | 77.1% | T4.07 | 3.5% | 2.4% | 78.2% |
| 37 | 77.7% | T4.06 | 3.1% | 2.9% | 79.9% |
| 38 | 78.1% | T4.04 | 2.6% | 3.4% | 81.2% |
| 39 | 78.0% | T4.04 | 2.3% | 3.9% | 82.5% |
| 40 | 77.9% | T4.05 | 1.9% | 4.6% | 83.5% |
| Lands | Commander on curve | Avg. commander turn | Screwed | Flooded | Kept first 7 |
|---|---|---|---|---|---|
| 32 | 48.2% | T5.23 | 8.9% | 0.5% | 71.3% |
| 33 | 49.6% | T5.18 | 7.8% | 0.6% | 73.4% |
| 34 | 51.0% | T5.14 | 7.0% | 0.8% | 75.1% |
| 35 | 52.2% | T5.11 | 6.3% | 1.0% | 76.6% |
| 36 | 53.3% | T5.07 | 5.5% | 1.2% | 78.2% |
| 37 | 54.5% | T5.04 | 5.0% | 1.4% | 79.9% |
| 38 | 55.9% | T5.00 | 4.3% | 1.7% | 81.2% |
| 39 | 55.8% | T4.99 | 3.8% | 2.1% | 82.5% |
| 40 | 57.3% | T4.94 | 3.4% | 2.4% | 83.5% |
| Lands | Commander on curve | Avg. commander turn | Screwed | Flooded | Kept first 7 |
|---|---|---|---|---|---|
| 32 | 76.8% | T4.78 | 5.6% | 1.0% | 71.3% |
| 33 | 78.2% | T4.74 | 5.0% | 1.3% | 73.4% |
| 34 | 80.0% | T4.68 | 4.4% | 1.5% | 75.1% |
| 35 | 81.8% | T4.62 | 3.9% | 1.7% | 76.6% |
| 36 | 83.7% | T4.57 | 3.4% | 2.1% | 78.2% |
| 37 | 85.2% | T4.53 | 3.0% | 2.5% | 79.9% |
| 38 | 87.0% | T4.47 | 2.5% | 2.9% | 81.2% |
| 39 | 88.2% | T4.42 | 2.2% | 3.4% | 82.5% |
| 40 | 89.3% | T4.39 | 1.9% | 4.0% | 83.5% |
Sources
- Frank Karsten, How Many Lands Do You Need in Your Deck? An Updated Analysis, TCGplayer.
- Simulations: FreeMully, Karsten keep rule with a free first mulligan, draw on turn 1, 10 turns per game.