Introduction
Skill DCs are where PF2e stops being just math and starts being table judgment. The rules give you anchors: simple DCs, level-based DCs, skill DCs, DC adjustments, secret checks, Aid, and proficiency gates. Somebody still has to decide which anchor fits the fiction.
This guide works from both sides of the screen. If you are a player, the first half tells you what number you are rolling against and why, so you can decide whether training, tools, time, or a better plan is the thing to change. If you are running the game, the workflow, adjustment, secret check, group check, and subsystem sections are the procedure for setting that number in the first place.
The goal is not to make every door, rumor, cliff, treaty, lock, and soup recipe scale to the party. The goal is to choose a DC that makes training, tools, time, risk, and preparation matter.
The Core Question
Before setting a DC, answer one question: what kind of obstacle is this?
| DC Tool | When To Use It |
|---|---|
| Simple DC ● | The task is defined by ordinary training: climb a rough wall, decipher a routine document, repair common gear, remember general lore. |
| Level-Based DC ● | The task is tied to a leveled challenge: monsters, hazards, spells, rituals, magic items, elite locks, curses, or adventure-scale obstacles. |
| Opposed DC ● | One creature acts against another creature's statistic, such as a skill DC, Perception DC, Fortitude DC, Reflex DC, or Will DC. |
| Subsystem ● | The scene needs several rolls and accumulating progress: research, influence, infiltration, chases, reputation. |
| No Roll ● | Success is obvious, failure is impossible, or the die roll would add no meaningful consequence. |
| Hard No ● | The character lacks the required tools, time, access, language, training, or physical possibility. Use sparingly and explain the fictional blocker. |
Hot Take: "roll and see" is not harmless. A DC is a promise about how hard the world is right now and what the party risks by touching the problem. If you cannot say that promise out loud in one sentence of fiction, the number came from panic wearing a fake mustache, and the check is just dice confetti.
The GM Workflow
The number matters, but the procedure matters more. A perfect DC with a dead-end failure state is still a bad ruling. Run the scene through these seven questions in order and the number usually picks itself.
| Step | Ask this | Good answer |
|---|---|---|
| 1 | Does this need a roll? | Roll only when failure, cost, time, or uncertainty matters. |
| 2 | What is the DC family? | Simple DC, level-based DC, opposed DC, or subsystem. |
| 3 | What changes the baseline? | Tools, time, environment, rarity, risk, pressure, help, or bad information. |
| 4 | Who rolls? | One lead, a group check, a secret check, or a subsystem turn. |
| 5 | What does failure do? | Cost time, reveal danger, attract attention, spend resources, worsen position, or give partial information. |
| 6 | Can players improve the odds? | Better approach, proper tools, Aid, magic, research, a bargaining chip, or more time. |
| 7 | What does success unlock? | Information, access, position, safety, progress, trust, treasure, or tactical advantage. |
Steps 2 and 3 are in that order for a reason. Choose the baseline family first, then adjust for conditions. Reversing them produces numbers that drift with the GM's mood instead of the fiction.
Simple DCs
Use simple DCs when proficiency expectation is the real question.
| Task Expectation | Simple DC | Best Use |
|---|---|---|
| Untrained ● | 10 | Common tasks most adventurers could attempt with basic competence. |
| Trained ● | 15 | Professional basics: routine medicine, common crafting, standard social procedure, ordinary survival. |
| Expert ● | 20 | Serious specialist work under normal conditions. |
| Master ● | 30 | Impressive work that should feel out of reach for casual dabblers. |
| Legendary ● | 40 | Mythic-level skill expression. Do not use this for a normal locked pantry unless the pantry is having a very dramatic week. |
Two things about this table get misread constantly. First, the rank you pick describes the task, not the character; GM Core is explicit that the matching rank is usually not required to succeed. Estimating that only a master of a niche Lore would know a piece of buried history gives you DC 30, and an expert can still roll for it and get lucky. Second, the gaps are huge. Expert 20 jumps straight to master 30 with nothing between. That coarseness is why GM Core steers hazards and combat-stakes checks toward level-based DCs instead: when someone's life is on the line, you want gradation.
Simple DCs are good for skills, equipment, actions and activities, and downtime tasks where the obstacle is not a leveled enemy or hazard.
Level-Based DCs
Use level-based DCs when the obstacle belongs to leveled game math.
| Challenge | Why Level-Based DCs Fit |
|---|---|
| Monsters ● | Creature abilities, Recall Knowledge, weaknesses, traits, and elite tactics should reflect the creature's level. |
| Hazards ● | Traps and haunts are encounter elements, not ordinary furniture with an attitude problem. |
| Spells And Magic ● | Identify Magic, checks and DCs, rituals, curses, and magic item mysteries often care about level. |
| Adventure Obstacles ● | A fortress gate, planar lock, royal forgery, or ancient ward can be level-tuned when it is part of the adventure challenge. |
GM Core's DCs by Level table runs from level 0 up to level 25. For a rough sense of the curve, a level 1 challenge sits at 15 and a level 20 challenge at 40, with the numbers climbing a little faster than one per level. Spells use a separate ten-rank column because spell ranks only go to 10th.
The same wall can legitimately take either treatment. GM Core's own example is a smooth metal wall in a high-level dungeon: call it a master-only climb at simple DC 30, or tie it to the 15th-level villain who built it and use the 15th-level DC. Both are called reasonable. What is not reasonable is making every ordinary task level-based. A level 15 character should not need mythic effort to climb a mundane fence. If the fence is the enchanted palisade of a demigod's prison, different story.
A character who invests in a skill eventually walks over DCs of their own level. That is intended. If you need a check that still threatens a high-level party, GM Core suggests reaching for the harder adjustments rather than inflating the level.
Running a Game This Week?
The GM Session Kit - a Session Zero template, a recap format, and a pricing cheat sheet. Free, and yours the moment you ask.
No spam, unsubscribe anytime.
DC Family Selection
GM Core hands the GM two baseline tools and one lever: simple DCs for tasks without a level, level-based DCs for challenges tied to a creature, hazard, item, spell, or level-scaled obstacle, and difficulty adjustments to account for unusual circumstances. Opposed checks and subsystems sit alongside those as the other two shapes a scene can take.
| DC family | Use it for | Do not use it for |
|---|---|---|
| Simple DC | Ordinary tasks keyed to training level: routine climbs, common repairs, local etiquette, practical survival, mundane locks, basic research. | Monsters, hazards, spells, items, rituals, or elite adventure threats with a level. |
| Level-based DC | Creatures, hazards, magic items, spells, rituals, curses, elite locks, major adventure obstacles, and level-tuned mysteries. | Everyday doors, ladders, normal gossip, or tasks that should stay mundane at high level. |
| Opposed DC | Contests against a creature's defenses, perception, deception, stealth, willpower, or active resistance. | Static objects that do not oppose the party. |
| Subsystem | Research, influence, infiltration, reputation, chases, victory-point scenes, and other structured sequences. | A single quick action that should resolve with one roll. |
"The party is level 12" is not a reason for a normal town door to become a legendary door. Sometimes the heroes are allowed to be competent.
Skill Modifier And Skill DC
Player Core gives the same practical structure for skill math:
| Number | Formula | Example Use |
|---|---|---|
| Skill Modifier | attribute modifier + proficiency bonus + bonuses - penalties | The number you add to a skill check. |
| Skill DC | 10 + skill modifier | The number another creature rolls against when testing your skill defensively. |
| Perception DC | 10 + Perception modifier | Commonly used against Stealth, deception, hidden threats, and observation. |
| Save DC | 10 + save modifier | Used by actions that target Fortitude DC, Reflex DC, or Will DC. See save progression. |
Opposed checks should not become opposed rolls. GM Core's general rule is blunt about it: when two sides are opposed, one rolls against the other's DC, and both sides do not roll. Initiative is the stated exception. That keeps the game from turning into a dice duel where probability wanders off wearing a novelty hat.
Proficiency Gates
Some skill actions require training. Others are untrained by default but may still require higher proficiency if the task is beyond casual knowledge.
| Gate | Use It For |
|---|---|
| Untrained Access ● | Basic attempts like common Recall Knowledge, climbing, swimming, hiding, or persuading in ordinary circumstances. |
| Trained Requirement ● | Specialized actions such as Decipher Writing, Identify Magic, Learn a Spell, Treat Wounds, Pick a Lock, or Disable a Device. |
| Higher Proficiency Requirement ● | Esoteric lore, masterwork crafting, legendary movement, hidden magical structure, or expert professional work. |
| Fake Gatekeeping ● | Blocking a reasonable attempt just because the GM wants the scene to go one way. Avoid. |
GM Core calls the mechanic minimum proficiency, and the wording is useful: a character below the required rank can attempt the check, but cannot succeed. The untrained barbarian is allowed to squint at the occult theorem. She just needs the chance to critically fail and walk away with confidently wrong information.
There is a sanity rail on this. GM Core advises that a task of 2nd level or lower should almost never require expert, 6th level or lower should almost never require master, and 14th level or lower should almost never require legendary. Otherwise no character of the appropriate level could get there at all.
The best proficiency gates clarify the world. The worst ones announce that the plot has locked the door from the outside.
Adjusting Difficulty
Adjust because the situation changed, not because the GM wants more suspense. There are two different levers here and mixing them up causes most of the confusion.
The first is the formal DC adjustment ladder, which GM Core applies to a baseline you have already chosen. It matters most for rarity and for skills of differing applicability.
| Relative difficulty | Change to the DC | Rarity it maps to |
|---|---|---|
| Incredibly easy ● | -10 | none |
| Very easy ● | -5 | none |
| Easy ● | -2 | none |
| Hard ● | +2 | uncommon |
| Very hard ● | +5 | rare |
| Incredibly hard ● | +10 | unique |
Those labels describe the task relative to its own baseline, not relative to the party. A very hard 2nd-level task is not very hard for a 10th-level PC. GM Core also notes that high-level groups often find "very hard" reads as standard, so plan accordingly when you want a real challenge.
The most useful application is the one people skip: different skills can face different DCs for the same object. GM Core's example is a 4th-level arcane tome at DC 19 for Arcana, very hard at 24 for someone reaching with Occultism, and easy or very easy at 17 or 14 for a character with the exact Dragon Lore that covers it. The adjustment reflects how applicable the skill is, and it sits on top of the character's own modifiers rather than replacing them.
The second lever is the quick improvised one. GM Core's general rules-of-thumb list says that if an effect raises or lowers the chance of success, grant a +1 circumstance bonus or a -1 circumstance penalty. Small, but PF2e's degrees of success bands are close enough together that a single point regularly moves a result across a line.
| Situation | Good adjustment |
|---|---|
| Excellent tools, strong plan, something to bargain with, accurate research | Easier DC, a circumstance bonus, safer failure, or faster progress. |
| Missing tools, bad lighting, hostile weather, pressure, damaged equipment | Harder DC, a circumstance penalty, more time, or worse position. |
| Extra time and no real danger | Consider skipping the roll or charging time instead. |
| Rushed work during combat | Require actions, use a harder DC, or limit attempts. |
| Uncommon, rare, or unique information | Apply the matching rarity step, require access, or use a narrower Lore skill. |
| Perfect plan that bypasses the obstacle | Do not roll. Let the plan work and move to the next consequence. |
| Impossible without a missing piece | Say what is missing instead of disguising "no" as a cruel DC. |
| Repeated attempt | Add time cost, resource cost, attention, escalating danger, or a changed approach requirement. |
Secret Checks
Secret checks protect uncertainty. The rule is mechanical, not a GM mood: if an action has the secret trait, the GM rolls it and tells the player the effect without revealing the result or the degree of success. Recall Knowledge, Decipher Writing, Identify Magic, and Seek for hidden things all carry it. The GM can decide to run some or all of those rolls in the open instead.
| Secret Check Use | Why It Works |
|---|---|
| Recall Knowledge ● | A critical failure hands the character incorrect information. That should be roleplayed, not solved by seeing a 3 on the die. |
| Search And Seek ● | Hidden threats stop being hidden if the player knows the roll failed. |
| Deception And Sense Motive ● | Social uncertainty is more interesting when characters work from observed behavior, not table metadata. |
| Public Roll Instead ● | Fine when suspense is lower, speed matters more, or the table prefers visible dice. |
Reporting the result is its own skill. Describe what the character perceives, not the die.
| Secret check type | Good result handling |
|---|---|
| Recall Knowledge | Answer the question truthfully on a success, add context or a follow-up on a critical success, and give plausible wrong information on a critical failure. Do not read out the stat block. |
| Sense hidden danger | Describe what the character notices, not the roll. |
| Detect a lie | Give behavioral tells, contradictions, or uncertainty instead of "they are lying." |
| Search for hazards | Reveal signs, suspicious details, safe routes, or where a closer look is needed. |
| Identify magic | Give tradition, function, danger, or an activation clue scaled to the degree of success. |
Keep PC modifiers on a card. GM Core suggests recording at minimum each player's Perception modifier, their saving throw modifiers with Will highlighted, and the skills that come up in secret checks. It prevents the sad little pause where everyone watches the GM hunt through notes while the table slowly loses oxygen.
Secret checks are not a license to be smug. The table should still understand what kind of action happened and why the result is hidden.
Aid, Follow The Expert, And Group Checks
When everyone wants to roll, decide how the group is actually cooperating.
| Method | Best Use |
|---|---|
| Lead Character + Aid ● | One expert leads, others help with specific preparation or positioning. |
| Follow the Expert ● | Exploration scenes where an expert guides allies through repeated checks. |
| One Roll For The Group ● | Group research, shared travel problems, or downtime tasks where one outcome fits the scene. |
| Everyone Rolls Freely ● | Usually bad for uncertainty. If five people roll until someone succeeds, the DC was mostly decoration. |
Aid has real numbers behind it. It is a reaction, the ally has to be willing, and the helper must have prepared during their own turn and explained to the GM exactly how they are helping. The typical Aid DC is 15, adjustable for unusually hard or easy tasks. A success grants a +1 circumstance bonus; a critical success grants +2, or +3 if the helper is a master with the check attempted and +4 if legendary. Critical failure hands the ally a -1 penalty. That downside is the reason "I help" is not a plan. Holding a rope, bracing a ladder, reading the second half of an inscription, distracting a guard, or checking a formula is a plan.
Follow the Expert is the exploration-mode answer and it is stronger than most tables realize. The ally being followed must be at least an expert in the skill and willing. While following, a character adds their level as a proficiency bonus even if they are untrained, plus a circumstance bonus of +2, +3, or +4 depending on whether the leader is expert, master, or legendary. That is how a party of non-climbers gets over the wall without five separate humiliations.
Group Checks And Dogpiling
| Problem | Ruling |
|---|---|
| Everyone wants to roll the same check | Pick one lead and allow Aid if others have a real method. |
| The whole party must sneak, climb, or endure | Use a group check or individual consequences. |
| One specialist should matter | Let the specialist lead, translate, examine, or guide. |
| The task takes time | One attempt may represent minutes, hours, or a scene, not one six-second button press. |
| Failure can be retried safely | Consider skipping the roll and charging time instead. |
| Failure can be retried dangerously | Each attempt should create noise, cost, exposure, damage, or clock progress. |
GM Core is direct about the math. When only one person has to succeed but everyone can attempt it, the sheer number of dice makes success near certain. If you want that task to carry real uncertainty, make it hard or very hard so that most of the party fails and the specialist is the one who gets there. That is the anti-dogpile rule in plain English: if five people roll for free, probability eats tension for breakfast.
Player Core Foundation
Player Core gives the player-facing checks and skills. Players choose training, attributes, feats, tools, spells, Aid, and tactics on the assumption that skill investment matters. If every task uses the same vague DC, proficiency choices feel fake.
| Player investment | GM responsibility |
|---|---|
| Skill training | Make trained, expert, master, and legendary choices visible in the fiction. |
| Skill feats | Let narrow feats bypass hassle, reduce risk, save time, or open special approaches. |
| Tools | Make the right kit matter, especially for Crafting, Thievery, and Medicine. |
| Spells | Let utility magic solve problems, then create new costs through time, noise, attention, or resources. |
| Aid | Let teamwork matter, but require a real helping method. "I also try" is not always Aid. |
| Exploration activities | Let initiative and exploration choices affect what rolls are even available. |
Player Core 2 Skill Pressure
Player Core 2 adds classes, ancestries, backgrounds, feats, archetypes, and items that push skill DCs constantly. Investigator turns information into action economy. Swashbuckler turns skill actions into combat momentum. Alchemist cares about Crafting, item identification, and preparation. The newer ancestry options push terrain, senses, movement, social pressure, and creature knowledge, which makes Survival, Stealth, Society, Nature, and Lore more than downtime decoration.
| Player Core 2 pressure | DC lesson |
|---|---|
| Investigator clue play | Do not make every clue a binary gate. Give partial information, leads, and sharper questions. |
| Swashbuckler skill-forward combat | Know when the skill action belongs in encounter mode and when the scene needs normal social pressure. |
| Alchemist item prep | Proper items should change risk, time, access, or difficulty. |
| Champion oath and deity pressure | Social and religious DCs should reflect public authority, doctrine, and consequence. |
| Oracle mystery play | Strange omens should point to meaningful checks, not vague vibes with dice attached. |
| Witch patron and familiar hooks | Familiar scouting, occult bargains, and hidden lessons need fair information boundaries. |
| Barbarian and monk positioning | Athletics, forced movement, and battlefield control all meet DCs. |
| New ancestries | Movement, senses, terrain, social rarity, and cultural knowledge can justify different DCs or different skills. |
Failure States
Before a roll, decide what failure does. GM Core's advice is to look for a way the party can fail forward, so the story moves with a negative consequence instead of stopping. It also tells GMs not to strain for it every single time; when nothing obvious presents itself, move on.
| Failure state | Use it when |
|---|---|
| Time cost ● | Travel, research, crafting, searching, social legwork, and pressure clocks. |
| Noise or exposure ● | Stealth, Thievery, locks, forced entry, scouting, and infiltration. |
| Resource cost ● | Tools break, materials are spent, spell slots matter, favors and money can absorb the setback. |
| Worse position ● | The party succeeds, but starts the next scene exposed, split, delayed, or noticed. |
| Partial information ● | Recall Knowledge, Investigate, social clues, mysteries, and lore scenes. |
| Closed path, open path ● | One route fails and another appears with a different cost. |
| Triggered hazard ● | The check was tied to a trap, cursed site, unstable machine, or dangerous magic. |
| Dead-End Nothing ● | Avoid when the check gates the only path. "You fail, I guess go home" is not drama; it is a scheduling problem. |
Failing forward does not mean failure is fake. It means the story keeps moving with consequences attached. "Nothing happens" is sometimes correct, but it should not be the house specialty.
Skill By Skill DC Thinking
The same skill can carry wildly different DCs depending on what the obstacle actually is. These are the levers worth asking about before you pick a number.
| Skill | DC thinking |
|---|---|
| Acrobatics | Terrain, balance, speed, armor, weather, and falling consequences matter. |
| Athletics | The obstacle's physical quality matters; not every wall scales with party level. |
| Crafting | Item level sets the DC, rarity adjusts it, and a repeat build can earn the easy step. Tools, workshop access, formulas, and time are the other levers. |
| Deception | Audience suspicion, proof, stakes, disguise quality, and prior reputation matter. |
| Diplomacy | Start attitude, what the PC has to trade, public pressure, and consequences matter more than speech length. |
| Intimidation | Fear has consequences; bullying every shopkeeper should change reputation. |
| Medicine | Training, tools, time, pressure, disease, poison, and battlefield danger matter. |
| Occultism | Weird lore should reveal usable angles, not just "it is weird." |
| Society | Institutions, law, etiquette, coded behavior, and city access drive the DC. |
| Stealth | Cover, senses, lighting, movement, noise, distance, and enemy alertness matter. |
| Survival | Weather, terrain, tracks, supplies, and creature behavior matter. |
| Thievery | Lock quality, traps, tools, time, noise, and danger determine the real cost. |
Earn Income deserves a separate note because its DC comes from a task level you choose rather than from the character. GM Core ties the highest task level available to the settlement's level, roughly 0 to 1 for a village, 2 to 4 for a town, and 5 to 7 for a city, with higher-level work concentrated in metropolises. A port town might still offer high-level Sailing Lore work. Nobody in most settlements is clamoring for Troll Lore.
Subsystems And DCs
Use subsystems when a scene needs multiple rolls and accumulating progress: research, influence, infiltration, chases, reputation, rituals, war turns, or complex social pressure. A subsystem can use DCs without letting one lucky roll solve the entire situation.
| Scene | Why a subsystem helps |
|---|---|
| Researching a cult | Multiple skills can reveal different truths. |
| Infiltrating a fortress | Failure can raise alert instead of ending the scene. |
| Winning a council | Different NPCs can care about different arguments. |
| Escaping pursuit | Athletics, Acrobatics, Survival, Stealth, and clever resources can all matter. |
| Running a ritual | Progress, backlash, assistance, and interruption can coexist. |
Subsystem DCs are built the same way as any other. GM Core's infiltration rules, for instance, tell you to start from the standard DC for the party level and apply the adjustment ladder, then vary difficulty by skill: low for Stealth while scouting unnoticed, hard for Perception because the enemy is also watching, very hard for Society because walking in like you belong needs an invitation or the right uniform. Spreading a scene across several skills at different DCs also means you are already prepared when the players improvise.
Fair DC Practice
Tell players enough to make informed choices. They do not need the exact number every time, but they should understand whether the task is routine, difficult, risky, or nearly impossible. If failure changes the story, say what kind of consequence is on the table before the roll.
| Say this out loud | Why it helps |
|---|---|
| "This looks routine for someone trained." | Signals that proficiency matters without naming a number. |
| "You can try, but without tools it will take longer and be harder." | Makes gear meaningful without hard-stopping creativity. |
| "Failure means time passes and someone notices." | Turns the roll into a decision. |
| "This is tied to the creature's level." | Explains why a challenge is harder than mundane precedent. |
| "One person should lead; others can Aid if they describe how." | Stops the everyone-rolls-until-success parade. |
| "This will be secret because the character would not know if they were wrong." | Keeps hidden information fair and transparent. |
DCs And Player Trust
Players do not need every DC in advance, but they do need to trust that the number follows the fiction. If a task is hard because the lock is masterwork, say that. If it is hard because time is short, say that. If it is hard because the party lacks the right tool, say that. If a player offers a better approach, reward it with a different skill, an easier DC, more information, or a safer consequence.
The strongest rulings are consistent, not identical. Two locks can have different DCs if one is a cheap storeroom lock and the other is a guild vault. Two speeches can have different DCs if one is given to friendly villagers and the other to a terrified council with assassins in the room. Explain the difference in-world and players usually accept the number.
Quick Reference
| Question | Reach for |
|---|---|
| Unleveled task, everyday obstacle, quick adjudication | Simple DC |
| Creature, hazard, item, spell, ritual, or level-scaled challenge | Level-based DC |
| Circumstances make the same task easier or harder | Adjustment (-10 to +10) |
| The subject is uncommon, rare, or unique | +2, +5, or +10 |
| Knowledge, detection, or hidden information | Secret check |
| Someone is actively resisting or hiding the truth | Opposed DC, one roll against the other's DC |
| Everyone must participate and the weakest link matters | Group check, made hard or very hard |
| One character leads and others have concrete ways to help | Aid at DC 15 |
| An expert guides allies through repeated exploration checks | Follow the Expert |
| Repeated rolls need progress, stakes, and a clock | Subsystem |
Common Mistakes
Scaling ordinary tasks to party level. A normal ladder does not become legendary because legendary heroes climbed into town.
Letting every player roll until success appears. Use one lead roll, Aid, Follow the Expert, a group check at a harder DC, or a time cost.
Adjusting before choosing a baseline. Pick the family, then apply the adjustment. Reversing that produces numbers nobody can defend later.
Using secret checks as a punishment. Secret checks protect uncertainty. They are not a GM power pose.
Ignoring tools and access. A trained thief without thieves' tools has a problem. A scholar without language access has a different problem.
Making rarity invisible. Rare facts should feel rare, and the rules give you the exact step to charge for it.
Forgetting skill DCs. Many actions target a creature's DC instead of making that creature roll. This matters for combat maneuvers, social pressure, and stealth play.
Using an impossible DC as a soft no. Say no, say what is missing, or offer a costly route.
Calling for rolls with no consequence. If failure changes nothing, skip the roll or change the scene.
Punishing creative solutions. If a clever plan changes the problem, change the DC or remove the roll.
Best Links
Use checks and DCs, proficiency ranks, skills, best skill feats, the GM workflow above, actions and activities, detection and senses, conditions, subsystems, social encounters, hazards, environment, initiative and exploration, GM screen reference, combat, and classes.
Return to the PF2e rules guide, the GM guides, or the PF2e guide hub.
