Creating an encounter often plays out the same for me: I come up with ideas for making the experience epic, find myself unable to solve the "what if they fail" problem, and then minimize or eliminate the cool ideas, leaving me with a feeling of disappointment.

Example 1: The PCs are travelling along an underground path when they come upon a massive cavern that is mostly a bottomless pit. Upon a stony island in the middle of this abyssal sea sits an extinct subterranean civilization's capital. It is a walled city, with a single bridge leading in and out. The ancient bridge that crosses the chasm is rickety and might teeter or break as the players traverse--no wait, they would just die if the bridge broke or they fell off. Ok, the bridge is just normal.

Example 2: The PCs are raiding a magical weapons factory that uses materia-like components that are highly reactive. There is a big showdown with the security team. The players have to be careful not to cause a chain reaction that--no wait, that would kill them all if the whole place exploded. Ok, there is no risk of a chain reaction.

As a result, the players have no real reason to fear anything except the bad guys. I feel like saying "Oops, you fell and you're dead and that's that" would be really unsatisfying for the players, but running around a world covered in bumper pads is unsatisfying too. How do I make a fantasy world deadly, but not so deadly that the players choose to stay home and work the family farm?

Group Checks, Success with a Cost, and Skill Challenges

These three concepts are meant to be used in tandem, but can be cherry picked to suit your style of play. I have the most experience using them together.

Group Checks

This is pretty simple. Half the group has to pass at doing some thing. In the examples you provided, this represents the party collectively crossing the rickety bridge, or a group effort to handle Materia.

Perhaps the Bare Minimum of 2 out of 4 people make the check/save.

"Wizardface nearly falls through a gap in the rickety bridge, but McFighterson manages to grab his wrist in time. The rest of the walk is slower as you take your time, but there are no other events of note."

You can secretly game this by knowing how likely it is for at least half the party to pass, based on the dc and their bonuses, and you don't have to worry about the wizard with the -1 acrobatics (or whatever).

If your group is bad at remembering inspiration, this is a good time, in my opinion, to remind them it exists.

Success with a Cost

"SwaC" is a hugely useful tool (okay, no one calls it that... yet). Maybe you want to leverage inspiration, mundane gear, spell slots, or hp.

Wizardface starts to slip through one of the tiles. McFighterson and La' Rouge are close enough to help, but they'd have to sacrifice the --insert thing here-- to get to him in time. Maybe it's not so bad at the bottom? What do you do.

This can be preemptively used, without rolling. I've had situations where I allow things to happen, wherein the party pays up front (usually at the party's suggestion).

Party: If we spend 2 inspiration, can we assume that the party makes it across?

DM: Of course. That sounds reasonable. Please describe your crossing and a hardship that you overcame

This allows them to enjoy the narrative feel where there is still a cost.

Skill Challenges

Disclaimer: I have adopted this from Matt Colville and his video on Skill Challenges.

This does not work for every scenario, but can be made to work for many. Choose a threshold, based on difficulty of the thing. Maybe it's 3 successes and 3 failures (adjust the number of successes for more difficult things). It may help you to think of this as a "race against time". In the case of the Materia example:

Barry Barbarian knocks over a box of the unstable element and one of the canisters begins to hiss and glow. You guys don't have long before it ignites and sets off the nearby canisters. The goal is three successes before three failures.

Then the group goes around, requesting to make proficient skill checks. The same skill cannot be used more than once, by the same character. Characters must be proficient in the skills. Players should be able to justify the skill in question. "No, unless you can explain how animal handling would help, you can't use it."

This might go something like the following:

Barry Barbarian succeeds at a survival check to retrace their steps.

McFighterson suggests using athletics to pry open a door that had warped from the reacting Materia. Success!

Rapier la' Rouge attempts to find a shortcut using investigation, but is unable. A vent of plasma bursts from the rivets of a nearby wall, causing light burns.

Wizardface is trained in Arcana and thinks he can predict a reactive fissure in the nearby steel-work. Success!


The Party escapes, managing to salvage one container of the materia. A more careful extraction would have made them a fortune, but the group feels fortunate to have kept their lives.

Honorable Mention: Make failure less fatal

Is there something about your cool scenario that could save the failing character? Perhaps the ravine is sloped and leads off into some unseen area? Perhaps the highly reactive Materia irradiates (bestows a curse) onto the nearby characters? Maybe the collapsing building only does 6d6 damage instead of instant death?

This isn't always going to fit the tone you want, but often, you can find that the unknown is just as terrifying as the certainty of death.

Try adding warnings and incremental degrees of deadliness to your environmental hazards.

Instead of a single failure leading to instant death, give your players chances to realize the danger and deal with it. Giving your players more chances and information makes them responsible for their potential deaths instead of you.

In the example of the rickety bridge, maybe it's stable enough that nothing really happens when the 35 pound Halfling Monk runs across. When the unarmored Human Wizard crosses, maybe there's some creaking and shaking, but no checks required. But when the 250 pound Goliath Fighter in full plate mail takes that first step onto the bridge, let him know that the thing starts creaking and trembling. If he continues, have him make a series of checks (probably Acrobatics) to see if he can step gently enough to keep the bridge stable. Rather than a single failure leading to death, however, maybe the first failure simply causes a single plank to fall from the bridge. A second causes a whole to break open under the fighter, but he is able to catch the edge and pull himself up without making another check. Hopefully at this point, he'll realize that he should find another way across. Perhaps an ally could cast Levitate or Fly on him, place him on a Tenser's Floating Disk as they cross, or simply use Dimension Door if they're high enough level. Maybe he's going to have to take off his heavy armor to safely cross. But if he's reckless enough to continue on at that point and fails a third check, that's when he makes the plunge. By making sure your players are absolutely aware of the deadliness of what they're attempting through warnings and the obviously increasing danger of each failed check, you also give them the burden of responsibility. It's not very smart for a fully armored Goliath to try to cross a rickety bridge, but if the player makes that decision and you give them a few chances, that's on them and not you.

Now let's take a look at the Materia factory example. Let's say, for example, that casting spells that deal Lightning damage is the trigger that could potentially set off a chain reaction and blow the place sky-high. If the party Wizard casts Lightning Bolt or Witch Bolt on turn 1, don't just say: "the factory blows up, game over". Instead, maybe describe that the materials on the nearby conveyer belt begin to glow and spark, and the party can feel the ground shaking around them as the energy spreads to the rest of the factory. If they don't get the hint, the next time a Lightning spell is cast, maybe one of the nearby Materia weapons explodes, forcing one or more party members to make a Dex save or take some damage. On the third Lightning spell, several such explosions occur around the room, and they can be heard elsewhere in the factory. At this point, it should probably be pretty clear that 1: Lighting is triggering the reaction, and 2: the reactions are getting more volatile with each occurrence. With this information, the party should hopefully see that now's the time to stop casting Lightning spells.

tl;dr: As long as the party has enough information, making a suicidal decision is their responsibility.

They die

Assuming that you are playing in a game where player death and even TPKs are OK with the players then kill them. If the players choose to have their characters attempt something that would kill them if the fail then, if they fail, they die. Part of enabling player agency is enabling the part that leads to bad consequences for poor choices. Remember, it only kills you if you die.

Of course, wise and clever players (which is all there ever are) will take precautions to prevent fatal consequences and their characters certainly have the resources to do so.

Example 1: The PCs are travelling along an underground path when they come upon a massive cavern that is mostly a bottomless pit. Upon a stony island in the middle of this abyssal sea sits an extinct subterranean civilization's capital. It is a walled city, with a single bridge leading in and out. The ancient bridge that crosses the chasm is rickety and might teeter or break as the players traverse ...

• 
  Fly spell (or Levitate or Feather Fall)


• Flying mounts


• Roping themselves together


• 
  Mending the bridge


• etc.

Example 2: The PCs are raiding a magical weapons factory that uses materia-like components that are highly reactive. There is a big showdown with the security team. The players have to be careful not to cause a chain reaction that ...

• 
  Nope it


• Negotiate and point out that this is a bad place for a firefight


• 
  Teleport away


• Finish the sentence "chain reaction that ..." with something less consequential than "everybody dies" or, alternatively, has an escalating cost that might end with "everybody dies" but is more than one step from "everybody's OK".

Non-binary Choices

It seems like the main problem you are running into isn't that your hazards are too hazardous, but that you are limiting yourself on their consequences. Any time you come up with a situation and the result is "Then everyone dies and the game is over" you should really step back and think about how you can change that to keep the game going instead.

Rolling for a Reason

Your first example with the rickety bridge has a bit of a fundamental flaw: We don't normally make characters roll for things like walking. If you want that to be an actual hazard then you need to add a reason for the slow and steady route to not be an option.

Maybe they are being chased by enemies and can't afford to go slow without being caught. Maybe there are cave bats that use that area as a hunting ground, and the PCs have to move faster to avoid getting attacked and knocked off. Maybe the bridge isn't terribly sturdy and will start to collapse after a few rounds, and normal movement won't get them off of it in time.

Save or Die Isn't Fun

There is a reason that all of the Save or Die spells are a high-level. Generally speaking PCs should very rarely run into anything that can do that to them because rolling poorly and then dying is inherently sucky. You should avoid doing the same thing with your environmental hazards as well. Figure out something negative that can happen that isn't just death and have that ready for the worst case scenario. Remember, even the most specialized character still has a 5% chance to roll a nat 1 and do as poorly as possible on any roll.

For your long bridge scenario I would think about all of the ways someone could fall off without it ending in a splat. Ancient magic teleports them into the city after falling a certain distance (unknown to everyone until the PCs that made it across get to the landing pad, for maximum tension). Those giant cave bats don't just knock people off, they also catch them and carry them back to their nests to feed the young. You could even pull a big fakeout and say that the floor is only 20 feet down, and its all an optical illusion that makes it look like a bottomless pit.

Same thing goes for your weapons factory fight. You are already most of the way there with how that scenario plays out. The PCs know that they need to be careful otherwise they set off a chain reaction and blow the joint. Just don't make that chain reaction happen all at once in the case where they fail. Someone hits the wrong box with a stray fireball, an alarm goes off and things start to glow and hum ominously, and then the PCs have to make their way back out of the weapons lab before the whole thing blows. Suddenly you have gone from a combat encounter with a limitation to a timed escape sequence, which is still exciting and interesting and dangerous but not immediately lethal.

The fun part of being a DM is getting to flex your creativity and tell an engaging story. PCs can and should fail at things, and those failures should have consequences. But you are always the final arbiter of what those consequences are. If you don't think it is fun then you can always find another path to take your players down.

Avoid lack of choice and instant death

You always need to have some tension of course. Peril generates excitement. But the players must always be in control and feel they have a chance.

Describing a rickety bridge may be peril enough, but has little chance of collapsing just by walking across. But if they face a battle at the end and a player has a choice of being careful (half speed) or running and requiring an Acrobatics check to avoid a gap in the floor...And even a failed check would just require a Dexterity save to grab the rope railing (and you may want a contingency plan, in case the player is really unlucky, such as a flying monster waiting to grab them for dinner, which actually allows the chance for the player to fight back at its nest!).

Carrying dangerous components can add a gradual peril. Perhaps if these components are uncovered or damaged or get too close to each other there is an effect that is obviously progressively dangerous. Maybe they start glowing and a generate a field that causes anyone in 30' to take 1d6 damage per round and suffer progressive levels of exhaustion (Con saves to avoid). Players (and any enemies) will recognise the danger but also know how to prevent it (probably).