Act 3 is not where runs are won — it is where they are cashed in. By the time the third act opens, the drafting is mostly over: your deck either has a working win condition, a draw engine and a block plan, or it does not, and no single card reward in Act 3 will change that. What Act 3 actually tests is execution: routing toward the fights your deck can win, spending upgrades and gold where they count, and walking into a 500-HP boss with a plan instead of a hope. This guide covers what a finished deck looks like, how to play the act itself, and what changes at Ascension 10 when the spire puts two bosses in the final room.
Act 3 is a consistency check
Every act of Slay the Spire 2 asks a different question. Act 1 asks whether your deck functions at all. Act 2 asks whether it can survive real pressure. Act 3 asks whether it does the same thing every single time you draw — because the enemies are now strong enough that a deck which wins "when it draws well" loses when it does not.
That reframes your decisions for the whole act. New card picks are rarely about power; they are about consistency. A card that makes your good turns slightly better is worth less than one that makes your bad turns acceptable, because at this depth the bad turns are what kill you.
It also reframes skipping. The deck management guide makes the case for a lean deck; Act 3 is where that discipline pays out. Every reward you skip is a slightly higher chance your win condition shows up on the turn you need it.
What a finished deck looks like
Four components, all present, all reliable. A win condition: poison stacking, an exhaust loop, a Barricade-and-Body-Slam bank, Star payoffs, strength — whatever your run found, it should be able to end a 400-HP fight inside ten turns. A draw engine: enough card selection that the win condition shows up every fight, not most fights. A block plan that works without the win condition on the table — the first two turns of every boss happen before your setup. And energy: enough to actually play the first three components on the same turn.
The exact card count matters less than the density — a lean mid-teens deck with three engines beats a bloated deck with one, and if you are unsure whether your deck is done, that is usually the answer: it is done when adding a card would make it less consistent, not more powerful.
Use the builds library as the checklist: each finished example lists the core cards, and if your run has none of them by Act 3, the honest play is to protect what the deck does have and route carefully, rather than hoping one rare fixes everything.
Act 3 route, campfires and shops
Route for reliability, not reward. The ideal Act 3 path hits the shops and campfires you need, fights the elites your deck can actually beat, and leaves an exit before the boss. Count rest sites before the final stretch — walking into the last two floors without a campfire option is how healthy runs die to one bad shuffle.
Campfire calls get simpler late: upgrade the card you play every single fight — your main block card, your main damage piece — and rest when the boss is close and health is short. The upgrade or rest guide rule still applies, just with higher stakes on both sides.
Shops in Act 3 are for the last problems: a final card removal if a starter card is still polluting your draws, the one block or energy answer your deck is missing, or potions for the boss. Elite rewards are still excellent — often run-defining — but only a clean deck should take them; our elite fighting guide has the benchmarks.
The Act 3 boss — and the Double Boss at Ascension 10
The final bosses sit near 500 HP — Doormaker, the largest of the current roster, runs about 489, with slightly more at Ascension; live values are on the bosses page. That number is the design statement: you cannot burst it down, so the fight is a war of attrition your deck has to win every turn for eight-plus turns.
Prepare like this: know the boss's telegraphs before you enter (the Ascension boss guide lists the big ones), schedule your potions for specific turns instead of emergencies, and spend the first two turns setting up scaling — a 500-HP pool forgives two quiet turns and punishes ten of them.
At Ascension 10, the finale becomes two bosses back to back with no rest between. Everything above applies twice over: sustain matters more than burst, and a deck that merely wins the first fight often loses the second. If your climb stalls at A10, the fix is almost always more block consistency and one more removal, not more damage.
Closing habits that save runs
Count the fight before you play it. Against the final boss, run the simple math on turn one: at my current damage, what turn does this end, and can my block cover every turn before it? If the answer is turn twelve and your block plan covers eight, you know immediately to route for one more upgrade or rest.
Finally, know when a run is lost. Sometimes Act 3 opens with your deck simply short — missing AOE, missing scaling, health already gone. Playing it out teaches more than abandoning it: take the fights you can win, practice the boss, and bank the knowledge. Every great player has a folder full of lost Act 3 runs and a shorter list of things they would not repeat. And if the whole run went sideways early, our beginner mistakes guide usually names the real culprit — it is rarely the last fight.