How much removal should a Commander deck run?
The usual answer, eight to twelve, carries a habit over from 60-card games. Moving to Commander made the deck 65% bigger and tripled the number of opponents, and it's the second change that sets your removal count. It also changes what removal is for, from answering everything to choosing what to answer. Here is the math for both, computed exactly.
Removal answers a permanent, interaction answers a play
This page covers the cards that deal with something already on the battlefield: kill spells, exile effects, bounce and board wipes. Counterspells and protection, which answer a spell while it's being cast, are their own budget with their own page. The difference shows up in timing. A threat on the battlefield waits for you, so a kill spell can be cast whenever the mana is spare, and you never have to hold mana up for it on the off chance. That patience is why removal is the bigger of the two budgets in most decks below Bracket 4.
Three opponents, one removal suite
Start with the comparison everyone's instincts are built on. A 60-card deck with eight removal spells, playing one opponent, has seen 1.87 of them by turn 8 on average, and every one is aimed at that opponent. Put the same eight in a 99-card Commander deck and by turn 8 you've seen 1.13, to share between three opponents. That works out to 0.38 answers per opponent, against the 60-card deck's 1.87.
Closing that gap would take about 40 removal spells in the 99, which nobody runs and nobody should. So Commander removal can't work as coverage, the way it does in a duel. It works as triage. Here is what real counts deliver, on the play in a 99-card deck, using exact hypergeometrics (the same math as this site's calculators):
| Removal spells | ≥1 by turn 3 | ≥2 by turn 6 | ≥3 by turn 8 | Average seen by turn 8 |
|---|---|---|---|---|
| 6 | 44.4% | 15.4% | 3.5% | 0.85 |
| 8 | 54.7% | 24.9% | 8.3% | 1.13 |
| 10 | 63.3% | 34.8% | 14.8% | 1.41 |
| 12 | 70.4% | 44.5% | 22.6% | 1.70 |
| 15 | 78.7% | 57.6% | 35.7% | 2.12 |
The ≥3 column is the one to stare at. Holding one answer per opponent by turn 8 happens in 14.8% of games at ten removal spells, and fifteen only gets it to about a third. In a four-player game, removal is scarce from the opening hand on.
What to answer, when you can't answer everything
Scarcity means choosing, and the pod gives you a way to choose. Your opponents run removal too, and a threat that endangers the whole table is one they're also motivated to kill. Spend yours on three kinds of target: the threat aimed at you, the threat nobody else can answer (an enchantment at a table with no white or green, say), and the threat that wins the game if it survives another turn. Everything else can wait a turn for someone else's answer.
This is also where the format genuinely disagrees. Some players run fifteen or more removal spells and kill threats as they appear, on the theory that waiting for the table is how you lose to the player who was counting on you to act. Others run seven or eight and lean on the table harder. Which one is right depends on your seat. A deck that is usually the biggest threat at its table gets no help from the others, so it needs the higher count. A deck that sits back can let the table do some of its work.
Sweepers are how one card scales with the pod
A kill spell answers one permanent. A board wipe answers every opponent's board at once, which makes it the one card in this budget whose value grows with the number of players. It's also the card that hits your own board, so the right sweeper count depends on what your deck is trying to do. A deck that wants creatures on the battlefield wants few wipes, or one-sided ones. A control deck that plans to be the player with nothing to lose wants many.
Many means more than most lists assume. Here is the chance of holding a wipe by each turn:
| Board wipes | ≥1 by turn 5 | ≥1 by turn 6 | ≥1 by turn 8 |
|---|---|---|---|
| 1 | 11.1% | 12.1% | 14.1% |
| 2 | 21.1% | 22.9% | 26.4% |
| 3 | 30.0% | 32.4% | 37.0% |
| 4 | 38.1% | 40.9% | 46.2% |
| 6 | 51.6% | 54.9% | 61.0% |
A deck running two wipes “as insurance” has one by turn 6 in fewer than a quarter of games. That's fine for a creature deck, where the wipe is an emergency button. It isn't fine for a control deck whose plan is the wipe: holding one by turn 6 even half the time takes six. One-sided wipes let a creature deck run more without hitting itself. Kindred Dominance spares the creature type you choose, and an overloaded Cyclonic Rift bounces only what your opponents control, though it's a Game Changer and costs a Bracket 3 deck one of its three slots.
Count the answers you can cast on curve
A six-mana wipe and a one-mana kill spell are both “removal,” but they aren't interchangeable at the count level. The threats you most need to answer early, like a cheap commander that snowballs or a mana engine that lands on turn 2, need an answer you can cast on turn 2 or 3. Only your cheap removal can do that, so give it a count of its own:
| Removal at 2 mana or less | ≥1 by turn 2 | ≥1 by turn 3 | ≥1 by turn 4 |
|---|---|---|---|
| 2 | 15.6% | 17.4% | 19.3% |
| 4 | 29.0% | 32.1% | 35.1% |
| 6 | 40.5% | 44.4% | 48.1% |
| 8 | 50.4% | 54.7% | 58.7% |
A deck with ten removal spells, four of which cost two or less, has a cheap answer in hand by turn 3 in 32.1% of games, not the 63.3% its total suggests. The expensive end has the opposite problem. On lands alone, a 37-land deck makes its sixth land drop on turn 6 in only 25.6% of games, so a six-mana wipe tends to arrive after the board it was meant for. Swords to Plowshares, Infernal Grasp and Rapid Hybridization are what cheap looks like.
Flexible answers count for more
A kill spell that only hits creatures is dead against an enchantment engine, and a Commander table will show you every permanent type sooner or later. Beast Within, Generous Gift and Chaos Warp hit any permanent, at the price of leaving the opponent a 3/3 or rolling the dice on what replaces it. You can't know your table's mix in advance, so a few of these are cheap insurance. A suite made entirely of creature removal is a smaller suite than its number says.
The answer, by what your deck does
| Deck | Spot removal | Board wipes |
|---|---|---|
| Creatures or tokens, wins with a board | 8 to 10, most at 2 mana or less | 1 or 2, one-sided if you can |
| Midrange | 7 to 9 | 2 or 3 |
| Control | 6 to 8 | 4 to 6 |
For most decks the totals land inside the familiar eight to twelve, so the convention isn't wrong about the number. It's wrong about the reason. The count isn't set by deck size, and it can't buy coverage of three opponents. What it buys is a few early answers you can cast on curve, enough flexible ones that no permanent type beats you, and as many wipes as your plan for the board needs.
Check your deck's own numbers
Enter your spot removal, how much of it costs two mana or less, and your board wipes. It opens on eight spot removal spells, four of them cheap, and two wipes.
With 8 spot removal spells (4 of them cheap) and 2 board wipes, you can answer an early threat by turn 3 in about 32% of games, and you hold one answer per opponent by turn 8 in about 15%.
On the play, no mulligans. “Three answers” counts spot removal and wipes together: one per opponent in a four-player game.
FAQ
How much removal should a Commander deck run?
Most decks land between nine and twelve pieces in total, but the split matters more than the sum. A deck that wins with creatures wants 8 to 10 spot removal spells, mostly costing two mana or less, and only 1 or 2 board wipes. A control deck flips that, with 6 to 8 spot removal and 4 to 6 wipes, because for control the wipe is the plan rather than the emergency button.
How many board wipes should a Commander deck have?
It depends on whether your deck wants a board. Creature and token decks should run 1 or 2, one-sided if possible, midrange decks 2 or 3, and control decks 4 to 6. Two wipes sounds like plenty but puts one in your hand by turn 6 in only 22.9% of games on the play, so a deck that really plans around wiping needs closer to six.
Why do Commander decks need more removal than 60-card decks?
Because there are three opponents instead of one, while the deck is only 65% bigger. Eight removal spells in a 60-card deck give you about 1.87 answers for your single opponent by turn 8. The same eight in a 99-card deck give 1.13 answers spread across three opponents, and matching the 60-card coverage would take around 40. That is why Commander removal is about choosing targets rather than answering everything.
Is a board wipe better than spot removal in Commander?
Neither is better, because they do different jobs. Spot removal is cheap enough to answer an early threat on curve and leaves your own board alone. A wipe answers every opponent's board at once, which makes it the only removal that gets better as the pod gets bigger, but it hits your board too and usually costs four mana or more. Most decks want both, in a ratio set by how much they rely on their own creatures.
Does instant-speed removal count as interaction?
Count it as removal. Interaction on this site means cards that answer a spell or ability as it happens, like counterspells and protection. A kill spell that targets a creature is answering a permanent, even when you cast it in response to something, so it belongs in the removal count. Counting one card in both budgets is how a deck ends up short on both.