How the DMG's XP budgets, multipliers, and difficulty thresholds work in plain language, where the math goes blind, and the heuristics that fill the gap.
You built the fight by the book. Four third-level characters, a budget that came out just under “deadly,” an ogre and his goblin entourage arranged with care. Two rounds later the ogre is dead, no one dropped below half health, and your players are asking what’s behind the next door. Or the mirror version: the book swore the fight was “medium,” and somehow the cleric is making death saves while the wizard hides behind a crate with nothing left but a cantrip and regret.
Every GM has lived both stories. And the usual conclusion, that the encounter math is broken, is the wrong lesson to take from them. The math isn’t broken. It’s doing a narrow job well, and we keep asking it to do a wider one. If you understand exactly what the numbers measure, you can let them carry the arithmetic while you handle the parts no table ever put in a book.
The Dungeon Master’s Guide builds encounters on three ideas: thresholds, prices, and a multiplier.
Thresholds. Each character, at each level, has four XP thresholds: easy, medium, hard, deadly. A third-level character’s thresholds are 75, 150, 225, and 400. Your party’s thresholds are just the sum. Four third-level characters give you a party line at 300 (easy), 600 (medium), 900 (hard), and 1,600 (deadly).
Prices. Every monster has an XP value tied to its challenge rating. A goblin (CR 1/4) costs 50 XP. An ogre (CR 2) costs 450. The bestiary is a menu with prices printed on it.
The multiplier. Here’s the part people skip, and it’s the part doing the most work. Numbers are their own kind of dangerous, so the DMG inflates the effective cost of a fight based on how many monsters are in it: one monster counts at face value, a pair at 1.5x, three to six at 2x, seven to ten at 2.5x, and up from there. You budget with the adjusted total; you award the base XP.
Run it once and it stops being abstract. Six goblins against our four third-level heroes: 300 XP at list price, doubled to 600 by the multiplier. That’s exactly the party’s medium line. A solid warm-up fight. Now add the ogre. The base cost is 750, but you have seven monsters now, so the multiplier steps up to 2.5x and the adjusted total hits 1,875. You didn’t add a medium monster to a medium fight. You crossed the deadly line at 1,600 and kept walking, and most of that jump came from the multiplier bracket, not the ogre’s price tag.
One calibration note, because the words are doing specific work: “deadly” doesn’t mean TPK. It means someone could go down and the party will spend real resources surviving. “Easy” doesn’t mean boring; it means a speed bump that costs a few hit points and maybe a spell slot. The four labels describe drain, not drama. (The 2024 DMG folds all of this into a simpler per-character budget without the multipliers, but the shape of the job is the same: price the opposition, weigh it against the party.)
That’s the whole system. It’s a good system. It will reliably stop you from throwing a young dragon at four first-level characters by accident, and for a GM running their first campaign, that guardrail alone is worth the page it’s printed on.
The XP budget models two things: how big the party is and how tough the monsters are on paper. Everything else that decides a fight is invisible to it.
Action economy. The multiplier is a patch over a deeper truth: fights are won and lost in actions per round. A single CR 6 monster against four fifth-level characters looks respectable in the budget, but every round it takes one turn and eats four. It gets focus-fired, stunned, restrained, or feared, and the moment it loses its one action, it loses the round. This is why legendary actions exist: they’re the designers admitting the math undercounts solos. If your boss doesn’t have them, the budget is lying to you about how scary he is.
Party composition. Twelve zombies are a completely different fight for a party with a fireball than for a party without one. A group with no healer treats burst damage as an emergency; a group with a life cleric treats it as Tuesday. The thresholds assume a generic party, and no party is generic. Yours has a moon druid, and that changes what “hard” means.
Terrain and position. Six goblins in an open field are a medium fight. The same six goblins shooting from a ledge over a chokepoint, in darkness, with a retreat tunnel behind them, will make your players hate goblins for the rest of the campaign. The statblocks didn’t change. The map did, and the map is nowhere in the formula.
The day so far. The thresholds quietly assume an adventuring day: six to eight medium-to-hard encounters between long rests, with the party’s resources draining across all of them. Almost nobody plays that way. A “hard” fight against a fresh party with every spell slot intact plays like a medium one. The same fight as the third encounter of the day, hit dice spent and the paladin out of smites, plays like a funeral. Before you trust a difficulty label, ask what’s left in your party’s tank.
Your actual players. The book can’t know that your rogue’s player reads statblocks for fun, or that the party owns a wand it was never supposed to find, or that the whole table plans like a special-forces unit. Table skill and magic items bend every threshold, and only you know which direction.
None of this means abandoning the math. It means letting the math do the arithmetic while you apply judgment on top. A few rules of thumb that have earned their keep:
Count actions, both sides. Before anything else, tally attacks-per-round for the party and for the monsters. If the monsters have fewer than half the party’s actions, the fight is softer than the budget claims. Add bodies, or give the big one legendary actions, lair actions, or a second turn on a different initiative count.
Treat “deadly” as a floor for rested parties. A fresh party with full resources will chew through a by-the-book deadly encounter more often than not. If it’s the first fight of the day and you want it to matter, build past the line, and compensate with the next rule instead of with apology.
Build exit ramps, not fudged dice. The honest way to correct a fight in play is through the fiction, not the rolls. Monsters have morale: when the ogre drops, the goblins run. Reinforcements are a dial: that second wave you prepped can arrive in round three, or never. Design fights in waves and you can throttle difficulty mid-encounter without touching a single die.
Ask three questions before the session. What do the monsters want (few things want to fight to the death)? What does the terrain do (who does the map favor, and how would the monsters use it)? What has the party got left (slots, hit dice, that one consumable they’re saving)? Three honest answers will move your difficulty estimate more than any recalculation.
Watch the drain, not the drama. After the fight, look at what it cost: hit points, slots, potions. If your “hard” encounters keep costing nothing, the labels have drifted for your table, and you should recalibrate what you build against, not wait for the book to change.
Here’s the quiet cost of all this: the thresholds, the price list, and the multiplier brackets are exactly the kind of arithmetic humans do badly at 11 p.m. on a prep night. You recompute the budget every time you swap a monster, forget the multiplier stepped up when you added the seventh creature, and round in whatever direction your optimism points.
This is the half of the job worth handing off. ScryRPG’s encounter builder runs the DMG math live while you work: drag creatures in and the difficulty meter re-scores against your party’s real size and levels, so you watch “hard” tip into “deadly” the moment that third ogre lands, instead of discovering it mid-session. Want the dice to keep some mystery? Set a variant, 2d4 wolves instead of four, and the count rolls at deploy. Attach a loot table and the reward is waiting the moment the fight ends, with initiative rolled server-side when you deploy so nobody’s asking whether the GM’s d20 was feeling generous.
And the boundary is deliberate. ScryRPG preps and deploys the encounter. Your table runs the battle wherever it likes: theater of the mind, a VTT, or graph paper. We keep the campaign; the battlemap is yours.
The math was never supposed to run your fight. It was supposed to keep you from ruining it by accident. Let it do that job perfectly, every time, without a calculator, and spend the attention you get back on the questions only you can answer: what the monsters want, what the map does, and what your players have left in the tank.
Build a fight, watch the meter, and see what your prep nights feel like with the arithmetic handled. Start free, or argue encounter math with us on Discord.
The answer isn't a number. It's knowing what kind of power you're actually giving. A framework for evaluating magic items by impact, not rarity.
Why mystery items deserve a place in your game, and how to make the reveal land at the table.
Six fair ways to divide D&D treasure, from need before greed to a self-balancing loot auction, each tested on the same hoard so you can pick one tonight.