TLDR
- Start with a plan to win, not a pile of “synergy” (which is often code for “my deck doesn’t end games”).
- Build a boring-but-functional 99-card skeleton first: mana, draw, interaction, finishers.
- Use a default baseline (then tweak): 37 lands, 10 ramp, 10 draw, 10 interaction, 7 win cards.
- Make your mana base match reality: colors, curve, and how many taplands you can tolerate before you age visibly.
- Playtest fast, change one thing at a time, and stop “rebuilding” every week like it’s a personality trait.
The hook
How to Build a Commander Deck is mostly learning to resist two urges: (1) adding every cool card you own, and (2) shaving lands because you “never want to draw them.” Congratulations, you have now built a 100-card improv comedy routine where the punchline is “I kept a two-land hand and died.”
Let’s fix that with a framework that builds decks that actually play Magic, not just shuffle beautifully.
Commander deck rules in 60 seconds
Before you brew, make sure you’re building something legal (or at least illegal on purpose).
- 100 cards total, including your commander (so it’s commander + 99). If you use a companion, it’s effectively a 101st card.
- Singleton: no duplicates except basic lands (and a few weird cards that explicitly break the rule).
- Color identity matters: every card in your deck has to fit within your commander’s color identity (including mana symbols in rules text).
- You start at 40 life.
- Your commander starts in the command zone, and costs {2} more each time you cast it from there (commander tax).
- Commander damage: 21 combat damage from the same commander knocks a player out.
If you’re ever unsure, check the official Commander rules page first. It’s faster than arguing about it for 11 minutes and then “letting it slide this time.”
Step 1: Pick a win plan and a table plan
A Commander deck isn’t “a theme.” A Commander deck is a plan with a theme as flavoring. Ask yourself two questions:
How do I actually win?
Pick 1–2 primary win paths:
- Combat (go-wide tokens, big stompy, voltron)
- Combo (infinite, semi-infinite, deterministic loops)
- Value snowball (out-resource the table until your board becomes rude)
- Lock or prison (you can do it, but expect social consequences)
Then choose enablers and payoffs like an adult:
- Enablers = cards that set up your engine (ramp, token makers, sacrifice outlets)
- Payoffs = cards that convert your engine into a win (Craterhoof-style finishers, drain effects, combo pieces)
A common failure mode: 12 payoffs, 4 enablers. That’s not “high ceiling.” That’s “my deck needs therapy.”
What kind of game is this deck for?
Not every deck belongs at every table. Wizards has been pushing optional language like Commander Brackets to help pre-game matchmaking, and you can use any shared vocabulary your group likes (brackets, power levels, “vibes,” whatever works). The point is to avoid the classic Commander experience: someone says “it’s a 7,” and then someone else says “cool, turn three win.”
Here’s a simple Rule 0 script you can steal:
- Casual / mid-power: “This deck is trying to win through combat around turns 8–10, no infinites, a couple board wipes, and I’m not running fast mana beyond Sol Ring.”
- Combo-leaning: “This deck has a couple two-card combos, but I’m not tutoring aggressively unless the table is also doing that.”
- cEDH-ish: “This is built to win fast and interact on the stack. If you’re not on that plan too, I’ll switch decks.”
Here is another MTG review.
Step 2: How to Build a Commander Deck with the 99-card skeleton
This is the part most people skip because it’s not exciting. It’s also the part that determines whether you get to participate in the game.
Start here, then fill the rest with theme cards.
The baseline template
Use this as a default. Adjust after playtesting.
| Category | Baseline count | What it’s doing |
|---|---|---|
| Lands | 36–38 (start at 37) | Make your land drops. Cast spells. Live. |
| Ramp | 8–12 (start at 10) | Get ahead on mana and tempo. |
| Card draw / advantage | 8–12 (start at 10) | See more cards, find your pieces. |
| Interaction | 8–12 (start at 10) | Stop opponents, protect your plan. |
| Win conditions / finishers | 6–10 (start at 7) | End the game on purpose. |
| Flex / theme package | Whatever’s left | The fun stuff, in controlled doses. |
“But my commander already does X”
Great. Still don’t cut the entire category.
- If your commander draws cards, you can trim draw a bit.
- If your commander ramps, you can trim ramp a bit.
- If your commander is removal, you can trim interaction a bit.
But commanders get removed, stolen, exiled, turned into elk, and occasionally just ignored while someone else wins. Keep the floor under your deck.
Quick power-level adjustments
Use these as rules of thumb, not commandments.
- Battlecruiser (big spells, longer games): +1 land, +1 ramp, +1 board wipe, +1 finisher
- Optimized mid-power: keep baseline, tighten curve, improve card quality
- cEDH-ish: fewer expensive wipes, more cheap interaction, more tutors, lower average mana value, faster mana (where your meta allows)
Step 3: Mana base and curve (yes, you have to do this part)
Commander games feel “swingy” when your deck is inconsistent. Most inconsistency is just mana problems wearing a trench coat.
Start with lands, then earn the right to cut them
A clean default is 37 lands.
Add lands if:
- Your commander costs 5+ mana
- Your deck’s average mana value is higher
- You’re running lots of expensive cards “because they’re cool” (they are, but they’re also heavy)
Cut a land only if you can explain why with numbers, not vibes:
- Very low curve
- A real ramp package
- Actual testing (not just optimism)
Watch your taplands
Tapped lands are fine in moderation. The trap is running so many that your deck is permanently one turn behind. If you keep thinking, “If only I had one more mana,” you probably did. It just entered tapped.
A practical rule:
- A few taplands are fine.
- A lot of taplands becomes a deck identity. And not a good one.
Color fixing is not optional
If you’re in 2+ colors, you need enough sources to cast your spells on time. The fastest way to check this is to throw your list into a deckbuilder (Moxfield, Archidekt, etc.) and look at color sources and curve distribution.
If you’re consistently missing a color early, your deck is telling you something. Believe it.
Step 4: Interaction that matches your meta (and your blood pressure)
A fun Commander deck still needs to interact, unless your plan is “hope everyone else plays badly.” Bold strategy.
Split your interaction into buckets
A simple baseline inside your 8–12 interaction slots:
- Targeted removal (6–8): Don’t make it all creature-only. Artifacts and enchantments are how Commander players commit crimes.
- Board wipes (2–3): These reset games you’re losing. Also, they punish the “I deployed 40 power and passed” lifestyle.
- Protection (2–4): Things that keep your engine alive (hexproof, indestructible, counterspells, fogs, whatever fits).
- One piece of graveyard hate: You don’t need to dedicate your deck to it. You just need something.
If your group is combo-heavy, shift interaction cheaper and include more stack interaction. If your group is creature piles, wipes and creature removal matter more. If your group is “we all ramp and nobody attacks,” bring a finisher and do everyone a favor.
Step 5: Put your deck on a turn plan
Mana curve is the spreadsheet view. Turn planning is the reality check.
Ask: what do you want to do on…
- Turn 1–2: ramp, setup, cheap draw, cheap interaction
- Turn 3–5: commander, engine pieces, value setup
- Turn 6+: convert advantage into a win, or at least a position that forces answers
If your hand is full of 6-drops and hope, you built a late-game deck without a way to reach the late game.
Step 6: Test, tune, repeat (without spiraling)
Here’s a low-drama tuning loop:
- Goldfish 5 opening hands.
Can you cast your early plays? Are you missing colors? Are you doing nothing until turn four? - Play real games, take tiny notes.
When you lose, write one sentence: “Lost because I ran out of cards,” or “couldn’t answer enchantments,” or “mana too slow.” - Change one thing at a time.
Swap a card, replay, learn. Don’t swap 17 cards and then wonder what caused the improvement (or the fire).
The goal is not perfection. The goal is consistency. Consistency is how your deck gets to “fun” more often.
Budget and proxies (the normal way, not the weird way)
Commander is expensive if you let it be. Proxies are a totally reasonable tool for playtesting and for keeping a casual group accessible, as long as you’re transparent and you’re not trying to pass anything as real.
Here’s a clean table script:
“Quick note, I’m running a few proxies for staples while I test the list. They’re clearly marked and just for gameplay. Everyone cool with that?”
If you want a deeper explainer on what proxies are (and what they are not), here’s a good one: What MTG proxy cards are and how to get them.
Also: sanctioned events are not the place for proxies. That’s not a moral statement, it’s just how organized play works.
Wrapping up
How to Build a Commander Deck comes down to one simple truth: your theme is only fun if your deck functions.
Build the boring skeleton first. Add the spicy theme package second. Then test, tune, and stop cutting lands because you’re bored of them. Lands are not boring. Not casting spells is boring.
FAQs
How many lands should I run in a Commander deck?
Start at 37. Go up if your curve is higher or your commander is expensive. Go down only if you have a low curve, real ramp, and you tested it.
How much ramp, draw, and removal do I need?
A strong baseline is 10 ramp, 10 draw, 10 interaction, then tune based on your commander and meta.
How many win conditions should a Commander deck have?
Aim for 6–10, depending on how your deck wins. Include enough that you see one in a normal game, and enough redundancy that one removal spell doesn’t delete your entire plan.
What if my commander draws cards or makes mana?
Trim that category a bit, but don’t delete it. Commanders get removed. Constantly.
Can I use proxies in Commander?
Often yes in casual play, if your group is good with it and you’re transparent. Usually no for sanctioned events. Ask first.
How do I tune a deck for cEDH?
Lower the curve, increase cheap interaction (especially on the stack), add more tutors and compact win lines, and build with the expectation that opponents will interact early and often.