Take ten cards out of a Commander preconstructed deck, put ten in, and stop. The ten that go in fill ten fixed roles, in this order: a mana rock, two land ramp spells, a draw engine, a draw burst, two pieces of removal, a board wipe, one recursion spell and one utility land. The ten that come out are the ten most expensive cards in the deck that do nothing the turn they arrive. On the 2026 precon worked through below, the ten cards cost 8.49 US dollars together and the deck's average mana value falls from 3.67 to 3.20.
The reason to cap it at ten is not budget. It is that a precon is already a legal, coherent deck, and the fastest way to make one worse is to replace forty cards with forty better cards that do not talk to each other. Ten is enough to fix the three things every precon is short of, and few enough that the deck still does the thing it was built to do.
Why ten changes is the whole job
Three construction rules make Commander upgrades different from upgrades in any other format.
A Commander deck contains exactly 100 cards including its commander (rule 903.5a). That is a hard count, not a minimum, so every card you add forces a card out. There is no "just add the good stuff".
Every card other than a basic land appears once (rule 903.5b). So you cannot add four copies of your best card, and a card that is good only when you draw it early is a card you will usually not draw at all. Consistency comes from having many cards that do a similar job, not many copies of one.
Every card has to fit inside your commander's colour identity (rule 903.5c), and colour identity counts every mana symbol in a card's rules text, not only its mana cost (rule 903.4). This is the rule that invalidates most upgrade lists you will find: a card can be the right effect at the right price and still be illegal in your deck. Check it before you buy, not after.
The practical consequence of all three: you are not building a new deck, you are swapping ten slots. How to play Commander covers the construction rules in full if any of that is new.
What is already in the box
Before buying anything, know what you already own. We counted the contents of the eleven 2026 Commander precons ranked in our precon guide, and every one of them contains Sol Ring, Arcane Signet and Command Tower. Ten of the eleven also carry Exotic Orchard, and nine carry Terramorphic Expanse.
Those are the five cards most first upgrade lists open with. If your deck is one of those eleven, you already own the first three and most likely the other two, so the money goes somewhere else. Every MTG Commander precon in 2026 has the full contents comparison.
The land count is the other thing already handled. Those eleven run between 37 and 40 lands, which is where a tuned deck lands too, so the mana base does not need rebuilding. How many lands go in a Commander deck? explains why that range is right.
The ten roles
Fill these in order. The examples are cards our data shows appear in a large share of decks for the commanders that can legally run them, and each is the cheapest printing our pricing feed carried on September 14, 2026, in US dollars. Prices move; the roles do not.
1. A mana rock in your colours. A two mana artifact that produces two colours fixes your mana and accelerates at the same time. The signet cycle is the budget answer: Golgari Signet and its nine siblings sit between 0.29 and 1.89 US dollars. The eleven 2026 precons ranked in our precon guide ship with two to eight rocks, and exactly one of them ships with fewer than four.
2 and 3. Two land ramp spells. A rock that dies to artifact removal is not a mana base. Rampant Growth at {1}{G} and Cultivate at {2}{G} put real lands onto the battlefield, which no one takes away. Outside green, the second slot becomes a second rock.
4. A draw engine. A permanent that draws you a card every turn, not once. Phyrexian Arena is the archetypal one at {1}{B}{B}: one extra card every upkeep for one life. This is the single most impactful slot on this list, because a precon's most common failure is running out of cards on turn eight.
5. A draw burst. Something cheap that refills a hand in one go. Sign in Blood draws two for {B}{B}.
6. Removal that is not restricted to creatures. Precons are heavy on creature removal and almost blind to everything else. Feed the Swarm at {1}{B} answers a creature or an enchantment, which in a format full of enchantment based engines matters more than the life you pay.
7. Catch all removal. One card that can answer literally any permanent, so no threat is unanswerable. Beast Within destroys any permanent for {2}{G} and hands back a 3/3.
8. A board wipe. A precon typically carries none, and a table with three other players will eventually build a board you cannot beat. Blasphemous Act reads {8}{R} on the card and costs {1} less for each creature on the battlefield, so in a real game of Commander it is usually a one or two mana sweeper. Its printed mana value stays 9 no matter what you pay, which matters for any effect that cares about mana value.
9. One recursion spell. Something that buys back the best card you have already lost. Eternal Witness returns any card from your graveyard to your hand for {1}{G}{G}, and it is a creature, so it also blocks.
10. One utility land. Not a dual land, a land that does a job. Rogue's Passage makes one creature unblockable, which is how a big creature actually finishes a game. This is the one slot where the card that goes out should be a basic land, so the land count does not move.
What to cut
Sort the deck by mana value, descending, and start at the top. A card at six mana or more has to do something the turn it arrives; a card that only sets up is a card you will never cast, because by the time you have six mana the game is already being decided.
The exception is worth stating clearly, because a rule applied blindly is how good decks get worse: keep the expensive cards that change the board immediately. A six mana creature that puts a board's worth of power onto the battlefield the moment it resolves earns its slot. A seven mana artifact that enters and then asks you to spend more mana activating it does not.
If cutting the top of the curve does not get you to ten, the next cuts are, in order: cards that need a board you do not reliably have, cards whose coloured mana requirements your mana base cannot support, and cards that only work with one specific other card in the deck.
A worked example: Blight Curse
Blight Curse is one of the two Lorwyn Eclipsed Commander decks, released January 23, 2026, led by Auntie Ool, Cursewretch in black, red and green. Out of the box it runs 39 lands, 27 creatures, four mana rocks, an average mana value of 3.67, and ten cards at mana value six or more. Only Dance of the Elements carries more.
The ten cards in
| Role | Card | Cost | Price, US dollars |
|---|---|---|---|
| Mana rock | Golgari Signet | {2} | 0.45 |
| Land ramp | Rampant Growth | {1}{G} | 0.33 |
| Land ramp | Cultivate | {2}{G} | 0.54 |
| Draw engine | Phyrexian Arena | {1}{B}{B} | 2.52 |
| Draw burst | Sign in Blood | {B}{B} | 0.63 |
| Removal | Feed the Swarm | {1}{B} | 0.23 |
| Catch all removal | Beast Within | {2}{G} | 0.89 |
| Board wipe | Blasphemous Act | {8}{R} | 1.06 |
| Recursion | Eternal Witness | {1}{G}{G} | 1.33 |
| Utility land | Rogue's Passage | land | 0.51 |
| Total | 8.49 |
None of those ten is in the deck already; we checked each against the printed list. All ten are inside Auntie Ool's black, red and green identity.
The ten cards out
Nine of the ten cuts are the deck's own top of curve, and the tenth is a single Swamp, because Rogue's Passage takes a land slot and the land count should stay at 39.
| Out | Cost | Why |
|---|---|---|
| Grim Poppet | {7} | Seven mana, then it asks for more mana to do anything |
| Liliana, Death Wielder | {5}{B}{B} | Seven mana for one counter a turn |
| Ferrafor, Young Yew | {6}{G} | Needs a board of counters it does not create fast enough |
| Midnight Banshee | {3}{B}{B}{B} | Triple black in a three colour deck, and it waits a full turn |
| Aberrant Return | {4}{B}{B} | Six mana that does nothing until graveyards are full |
| Carnifex Demon | {4}{B}{B} | Six mana, then pay again per counter |
| Chimil, the Inner Sun | {6} | Six mana and the payoff waits until your end step |
| The Reaper, King No More | {2/B}{2/R}{2/G} | Six generic mana for two counters on entry |
| Vraska, Betrayal's Sting | {4}{B}{B/P} | Six mana planeswalker that needs to survive a turn |
| One Swamp | land | Replaced by Rogue's Passage |
The card that stays: Grave Titan at {4}{B}{B} is the tenth card at six mana, and it is the one to keep. It is a 6/6 with deathtouch that brings two 2/2 Zombies with it, so it puts ten power across three bodies on the battlefield the moment it resolves, and four more every time it attacks. That is the difference this section is about.
What the swap did
| Before | After | |
|---|---|---|
| Lands | 39 | 39 |
| Cards at mana value six or more | 10 | 2 |
| Average mana value, nonland cards | 3.67 | 3.20 |
| Mana rocks | 4 | 5 |
| Dedicated land ramp spells | 0 | 2 |
Two cards at six mana or more remain: Grave Titan, kept deliberately, and Blasphemous Act, whose printed 9 is almost never what you pay.
Which precon to try this on next
The method is the same for every deck; only the ten cards change. Three that reward it most, in order:
- Blight Curse, the worked example above. Once the first ten has settled, the same method supports a second ten for a table that has already upgraded.
- Witherbloom Pestilence, the Secrets of Strixhaven black and green deck: it ships with the fewest mana rocks of the eleven, two, and the most creatures, 39, which is the widest gap between what a deck has and what it wants among the eleven 2026 precons ranked in our precon guide.
- Dance of the Elements, the five colour Lorwyn Eclipsed deck: the hardest of the eleven to upgrade well, because every slot is a mana base question first.
When ten swaps stop being enough and you want a deck of your own, how to build a Commander deck is the same thinking in seven steps, from a blank list.
Let the deck builder check the swap
Open the deck builder, no account needed, set your commander, and add the cards you are swapping in. The Validation tab checks all three construction rules as you edit: the exact 100 card count (rule 903.5a), one copy of each nonbasic card (rule 903.5b), and colour identity against your commander (rule 903.5c), which is the check that catches an upgrade list written for a different deck. It also flags anything on the Commander ban list; the ban list, explained has the current one.
If you would rather start from a description than a list, tell Grover, our AI deck assistant, in one sentence what you want the deck to do, and it drafts one for you. No account needed there either. How to build a Commander deck with an AI assistant shows what that looks like end to end.
Frequently asked questions
How do you upgrade a Commander precon?
Swap ten cards. Add a mana rock, two land ramp spells, a draw engine, a draw burst, two removal spells, a board wipe, one recursion spell and one utility land. Cut the nine most expensive cards that do nothing the turn they land, plus one basic land to make room for the utility land.
Is upgrading a precon worth it?
For ten cards, yes. On the worked example above, ten cards costing 8.49 US dollars in total took the deck from ten cards at six mana or more down to two, and added the two things precons are reliably short of, repeatable card draw and a board wipe. Beyond about twenty cards you are building a new deck, and starting from a precon stops being the cheap path.
Which Commander precons are the best to upgrade?
The ones with the clearest gap between what they ship with and what the deck wants. Among the 2026 releases, Witherbloom Pestilence ships with two mana rocks, and Blight Curse ships with ten cards at six mana or more. Both have an obvious ten.
What counts as an upgraded precon?
There is no official definition, because the ban list and the Commander brackets are format policy, not rules. Since the October 21, 2025 update, a precon is no longer the official reference point for any bracket, so judge the deck by its list, before the swap and after it. A ten card swap of this kind makes the deck more consistent without changing what it was built to do. The brackets conversation is about which cards you add, not how many.
Is 26 creatures enough for a Commander deck?
It is a normal number. The eleven 2026 precons ranked in our precon guide run between 19 and 39 creatures, and nine of them sit between 23 and 36. What matters more is whether the deck can keep drawing cards once the creatures are gone. How many creatures should a Commander deck have? goes further into it.
Is 32 lands enough for Commander?
It is below what each of the eleven 2026 precons ranked in our precon guide ships with; they run 37 to 40. Thirty two lands works only in a deck built deliberately around a low curve and extra ramp, and a precon is not that deck. Leave the land count alone unless you are also changing the curve.
Do I have to check colour identity on every upgrade?
Yes, and it is the check people skip. Colour identity counts mana symbols in a card's rules text as well as in its cost (rule 903.4), so a card that looks mono coloured can be illegal in your deck. Two of the 2026 precons are led by exactly this case. The Validation tab does the check for you.
Sources
- Mana Grove card database, MTGJSON build of September 14, 2026: the printed contents of the eleven 2026 Commander precon decks ranked in our precon guide, including the Blight Curse list used above; the land, creature, mana rock and mana value counts before and after the swap; and the presence of Sol Ring, Arcane Signet and Command Tower in every one of the eleven.
- Mana Grove pricing feed, provider prices as of September 14, 2026, cheapest available printing of each card in US dollars. Prices move daily; the figures above are a snapshot on that date and are informational only.
- Mana Grove commander staples dataset, snapshot of June 27, 2026: the share of decks per commander that run each card named in "The ten roles", used to pick the examples.
- Comprehensive Rules, effective August 7, 2026, rules 903.4, 903.5a, 903.5b and 903.5c: https://magic.wizards.com/en/rules. Read on September 19, 2026, from the snapshot our rules engine keeps. The 100 card count, the single copy requirement, the colour identity requirement and the rule that colour identity counts mana symbols in rules text.
The counts and prices above were measured on September 19, 2026 against the September 14, 2026 build, and what this guide says about the 2026 precons was brought up to date on October 24, 2026. The date at the top is the last revision.
