A painting of a boy holding a torch and a girl in a blue cloak.

Torchbearer is a great game. It makes the environment feel dangerous and ensures every stash of copper pieces feels like a wondrous treasure.  Torchbearer has even revitalized my interest in dungeons, something I thought was dead forever. But like any game I play a lot, I eventually find areas that could use improvement. After all, no rules system is perfect.

Fortunately, we can fix flaws through the magic of house ruling! I’ve made dozens of house rules for Torchbearer over the years. Most of them were to assist me during a specific campaign, but a few worked so well that I’ve used them over and over again. Since Torchbearer can be a challenging game, I wanted to share my most successful creations. I hope they’ll improve your campaigns as much as they have mine.

1. Relaxed Class Restrictions

The famous Liberty Leading the People painting.
You can take our lives, but you can never take our character creation choices!

Torchbearer is a tribute to old-school Dungeons and Dragons using modern game design techniques, so it contains numerous throwbacks to an earlier age of roleplaying. Most of them are harmless, but one sticks out: strict restrictions on what class you can choose based on your fantasy race, or stock, as Torchbearer calls it.

Humans can choose from a multitude of classes, but dwarves, elves, and halflings aren’t so lucky. They get one class each. Dwarves can only be adventurers,* elves can only be rangers, and halflings can only be burglars. While this might have made thematic sense in the first days of D&D, it doesn’t anymore. Now it just feels like a pointless restriction on what kind of character a player can make. It also feels like a weird bit of fantasy bigotry, with the implication that no species besides humans have any variety.

To fix all this, I always allow my players to mix and match their character’s stock with whatever class they like. Humans can be rangers, halflings can be warriors, and dwarves can be wizards. This gives the players more freedom and creates a deeper world in one fell swoop.

Mechanically, this is fairly simple. PCs use the Nature descriptors from their stock, but get everything else from their class. A halfling warrior would still have sneaking, riding, and merrymaking as their descriptors, but they’d get a warrior’s skills, attribute points, and class abilities. For non-humans taking a traditionally human class, I let them choose between their stock’s normal starting trait or the starting trait for their class. So our halfling could start with either Hidden Depths or Heart of Battle, depending on what the player was most interested in.

If a human is taking a traditionally non-human class, I usually substitute the class’s starting trait with one from the general list. For example, if a human PC was taking the ranger class, which is normally elf only, I’d substitute First Born for something like Sharp-Eyed. That way you don’t end up with PCs sporting traits that don’t thematically fit them.

Finally, I check the class abilities to make sure these changes haven’t upset anything. Usually, you only need to change the flavor text of certain abilities so they work with the new class/stock combo. The only exception is the ranger, which gets the choice between an outdoorsy-type ability and a magic spell each level. This is supposed to represent the inherently magical nature of elves, but it doesn’t really work for any non-elf rangers. Why would dwarves gain magic by ranging around the wilderness?

To fix this, I’ve created an alternate set of abilities for rangers to choose from. Feel free to use it, or craft your own!

Oren's Ranger Abilities

Level 2

  • Stone Thrower: A ranger never counts as unarmed provided they have a stone. A good stone acts as a weapon but grants no other bonus.


  • Wilder: Add +1 to the 2d6 Camp Events roll.

Level 3

  • Essence of the Earth: +1D to recover from exhaustion. Rangers are rugged folk.


  • Skirmisher: Improved leather armor. When you’re wearing leather armor, roll 2d6 to deflect a blow. If either die comes up a 4–6, you reduce the incoming thrust by -1s.

Level 4

  • Fearless: Rangers have seen the primal forces, and fear holds little meaning for them. When the afraid condition is handed out by the GM as the result of a fight, you do not mark a condition.


  • Hardy Stock: This ranger comes from hardy stock. Add +1D to recover from the sick condition or any tests to resist poison.

Level 5

  • Master: Double the benefits of using a bow, sword, or dagger in a fight: choose bow, sword, or dagger. A sword grants +2D per action; a bow gives +4D to maneuver or counts as a longer range than all other missiles; a knife grants two free disarms on a successful Maneuver.


  • Rain of Arrows: Once per conflict, after performing a successful Attack or Feint action, the ranger may permanently expend one slot of arrows to inflict one damage on all enemies.

2. Extra Traits for Reduced Nature

A mural showing people from around the world.
Pure Diversity by Mirta Toledo used under CC BY-SA 4.0

During Torchbearer character creation, there’s a section where you answer three questions about your character to determine what their starting Nature is. These questions help determine how closely your character adheres to the values they were raised with and how much they represent a “typical” example of their stock.

For example, the first dwarven Nature question reads:

If your kin are slain and their halls plundered, would you spend your life plotting and exacting revenge, or would you tally your losses and move on to greater challenges?

A character who seeks revenge is more typically dwarven, while a character who moves on is less connected with the traditional values of those who dig beneath the earth. Mechanically, that usually means a choice between increasing your character’s starting Nature or changing one of their traits to represent their unusual choices. In this case, a dwarf who doesn’t seek vengeance can change one of their starting traits to Honorable or Jaded.

While that’s very flavorful, it’s also a mechanical trap. Nature is a very powerful ability. It gives PCs major bonus dice when they spend meta currency, and can even be used as a substitute for skills the character doesn’t have. Traits, on the other hand, are mechanically identical. Changing a trait from one to another does nothing for the character other than altering the flavor of when each trait can be used.

That’s disappointing, especially for new players who don’t know any better. They take the time to answer the Nature questions based on what they think will make the best roleplaying, only to find out that they’re at a serious disadvantage with anyone who answered more tactically.

Punishing players for roleplaying is a special pet peeve of mine, so I always rule that characters get a new trait in addition to the traits they already have, rather than as a replacement. Now players are choosing between a higher Nature or more traits, a choice that’s actually balanced. Of course, this does mean that PCs can start with four traits rather than the normal maximum of two or three, but that’s not any more powerful than starting with a high Nature, so it doesn’t unbalance the game. It also gives the PCs more opportunities to use traits against themselves, a really cool aspect of Torchbearer’s rules that can be difficult for new players to grasp.

3. Skill Customization

A painting of George Washington and his family.
Everyone in the party starts with Cartographer!

Most of a Torchbearer PC’s skills come preset with their class. A warrior always starts with Fighter 4, Hunter 3, Commander 2, Mentor 2, and Rider 2. Meanwhile, a cleric always starts with Ritualist 4, Theologian 3, Fighter 2, Healer 2, and Scholar 2. The game has a couple of customization options available, but not many.

This system has a number of advantages, especially for new players. These preset skill packages help ensure that each party has at least some of the skills they’ll need to delve into deep dungeons, and it prevents analysis paralysis for neophytes who don’t know what any of these skills do anyway. Segregating skills by class also helps each class feel distinct and different, which is a valuable bit of flavor.

However, after you’ve played a few Torchbearer campaigns, the skill packages start to feel restrictive. Players who already know the game want to experiment, but that’s very difficult with such limited options. It might be fun to play a character who starts with a high Healer skill or a character with lots of social skills, but all such variation is hindered by the class-based skill packages.

On the bright side, it’s super easy to open up the skill list and let your players do as they like. This isn’t something I recommend with an inexperienced group, but if everyone at your table is an old hand at adventuring, then it can breathe some much-needed novelty back into the system.

The math is super simple. Each PC gets 3 skills at rank 3 and 3 skills at rank 2. They then get 3 points that can be used to either increase an existing skill, to a maximum of rank 4, or purchase a new skill at rank 2. Humans get an extra skill point to make up for having only one later on. This replaces the skills a character would normally get from their class, hometown, social graces, and specialty. PCs’ other abilities would still come from their class, which is why it’s important to make sure your players know what they’re doing, so they don’t accidentally create a warrior who can’t fight.

That’s all you need! Now you’re ready to bring custom skills to your game. But be warned: with great customizability comes great potential for imbalance. None of my players have yet found a game-breaking combination of skills, but that doesn’t mean there isn’t one out there.

4. Game Balance Tweaks

A set of balanced scales.

We’re finally out of character creation, which means it’s time to talk about everyone’s favorite design challenge: game balance! For the most part, Torchbearer scores pretty well in this department, especially compared to other dungeon-delving fantasy games. But no system is perfect, and over time I’ve noticed a few issues that need to be addressed. None of them are deal breakers on their own, but they add up.

Conflict Actions

If there’s one thing I routinely critique Torchbearer for, it’s the conflict actions. In short, there’s rarely a reason to do anything but choose Attack over and over again because Attack is the action that actually wins the conflict. There’s no counter to this strategy, and it makes conflict really uninteresting.

My solution is to give the Defend action a bonus against Attack. Since Feint already gets a bonus against Defend, and Attack gets a bonus against Feint, this creates a simple rock-paper-scissors mechanic that stops any one action from becoming too powerful. I’ve found that two bonus dice are usually enough, though your mileage may vary.

The Abstemious Ability

Abstemious is a level-two ability for the burglar class that allows the character to alleviate their hungry condition in exchange for gaining the angry condition. This is supposed to represent the PC tightening their belt and pushing through the hunger even though it makes them irritable.

The problem is that hungry is a much less severe condition than angry, and it’s much easier to get rid of. There are very few cases where a player would prefer angry over hungry, so the ability won’t see much use. That’s a major disappointment, since gaining new class abilities is often a highlight of the game.

My solution is to reverse the process. Once per session, this ability lets the burglar become hungry in order to erase angry. I also change the name to something like Eat Away Your Feelings, but that part is optional. This trade is actually advantageous for the character, so the player won’t feel ripped off with their new ability.


PCs make a lot of purchases when they come back to town, and it would be a pain to haggle over each one. Instead, Torchbearer allows each character to make one Haggler test at the start of a town phase, and if they roll well, they get a small bonus for their clever negotiating.

That sounds great until you do a little math and find out that the bonus is too small. It’s usually equivalent to one die of cash, and simply making the Haggler test increases the character’s lifestyle cost, which is the equivalent of two cash dice. That makes it nearly impossible to so much as break even, let alone come out ahead, and that’s assuming the PC rolls well. If they roll poorly, they’re penalized.

On the bright side, this trap has a simple solution: eliminate the lifestyle increase from haggling. It doesn’t really fit anyway, since it’s not clear how haggling makes the PC’s stay in town more expensive and eliminating it lets the player decide if the bonus from haggling is worth the risk of a possible failure, rather than being in the red no matter what they roll.

Finding Work

Finally, we have another town activity: getting a mundane job. This option is really useful when the party is down on its luck, as it allows PCs to earn a little cash so they can afford enough supplies to mount another expedition in search of treasure. At least, that’s the theory.

In reality, finding work is another option that fails due to basic accounting. Taking the “find work” option increases a character’s lifestyle cost by one, which as we’ve already covered, takes at least two dice of cash to pay off. But finding work only pays one die of cash in most cases.* You can see the problem here.

Contrary to what you might think, the solution is not to make finding work pay more. The moment finding work pays more than it costs, it becomes overpowered. Instead, my solution is to change the cost from money to a condition. In my games, finding work gives two dice of cash, at the expense of making the character leave town with the exhausted condition, representing how they were working a side job when they should have been resting. This allows the PC to make a tactical choice over which is more important: having a little extra cash, or being fully awake when the monsters show up.

5. Advancement From Helping

Hamlet gets all the helping dice in this scene.

Torchbearer’s rules for helping another character are fairly generous. If you have the same skill as another PC, you can give them an extra die in exchange for sharing the risk of failure. Multiple characters can do this at the same time, adding even more dice. The ease of helping is good for the game, as it rewards players who spread their skill points out since more skills mean more chances to help.

However, one area where the helping rules are significantly less generous is advancement. Torchbearer has a learn by doing system, where skills and abilities need to be used a certain number of times before they increase, but these rules only apply to the character actually making the roll. Helping characters get no advancement, even though they share the risk of failure.*

This usually ensures that players who spread their points out never get to advance their skills. They’re less likely to have the highest rank in something, and it usually works best if the PC with the highest rank rolls while everyone else helps. With the more specialized characters getting all the rolls, their skill level will continue to rise, creating even more pressure to let them handle the dice.

This just doesn’t feel good if you’re a PC with more spread-out skill points. Players like to see the numbers on their sheets go up, and it’s demoralizing to watch your skill stay the same while someone else’s increases.

My house rule is pretty basic: make helping count for advancement. It’s fairer to low-skilled characters, and it’s just more intuitive. Helping someone else through a task is experience too, and it feels like characters should benefit from that experience.

Be warned: PC skill levels go up a lot faster under these rules. However, that’s not as unbalancing as you might think. In most cases, the character with the highest skill is still making the roll with the same number of helping dice, so the party’s overall rolling power hasn’t gone up much. The main exception to this is combat, where everyone has to roll their own skills, and situations where the party is split up or otherwise unable to coordinate.

These scenarios will certainly require some adjustment on your part to make sure the party doesn’t have too easy a time of it, but I’ve found that trade-off is well worth the increased player enjoyment.

6. Simplified Resources

A painting of barrels full of money.

A huge portion of Torchbearer’s gameplay focuses on resources. The party must delve deep into ancient ruins in search of treasure, and they need food to eat and torches to light the way. When they come back to town, they’d better hope that their treasure is enough to replace all their supplies and pay for a healer to patch up their injuries.

This resource management is what makes Torchbearer stand out from its competition, but the actual Resources attribute is a bit of a problem. Resources is an abstract representation of a character’s material wealth, which is fine in theory but turns into a headache fast.

The first problem you’ll notice is that PCs often need to buy a lot of supplies and gear, which means they need to roll Resources over and over again. This is incredibly tiresome, but it gets worse when players try to figure out how cash dice work. You spend them to supplement a Resources roll, and they also insulate you from taxation, the main penalty of failing a Resources roll. Then you need to figure out the exact number of cash dice to spend so you’re safe from tax, but you can also succeed or fail the roll based on what you need for advancement… It’s a mess.

I’ve actually got multiple house rule options to deal with this problem. The first is essentially what I did for Rising Tide: lump supplies into a character’s lifestyle roll. That way players just make one big roll at the end of the town phase, rather than a dozen or more as they replace every piece of gear in their pack. You may wish to tweak the math a little, but I usually rule that increasing lifestyle cost by one buys a character:

  • Three slots of fresh rations, wine, candles, or torches in any combination.


  • Two slots of preserved rations or lantern oil in any combination.

You may notice that the first option handles the Ob 1 supplies, while the second is for Ob 2 supplies. Feel free to add more items to the list, if your players buy a lot of them.

To supplement this option, I also change Resources so that it only requires successful rolls to advance, but that it needs more of them.* Together, these two rules reduce the number of redundant Resources rolls at the table and eliminate the weird math players go through where they want to fail a resources test but not so badly that they’ll be taxed.

Another option is to do away with Resources rolls altogether. Instead, simply have your players buy gear and supplies with cash dice at a one-for-one basis. An Ob 2 piece of gear costs two cash dice, an Ob 3 piece costs three cash dice, etc. This is how the game is played early on anyway since PCs all start with a Resources of zero; you’ll just be extending it.

Not rolling to buy things speeds up the game a lot, and if you still want an attribute to represent the party’s slowly accumulating wealth, you can keep Resources around to represent the PCs’ credit and investments. Instead of rolling to buy something, players could roll their Resources at the start of each town phase, generating a die of cash for each success.* To advance it, players could invest a number of cash dice based on how many rolls it would take to raise the attribute in the base rule. The exact number is up to you, but I’ve found that three dice of cash for each roll works well.

7. Actions for Everyone

A painting of a town with lots of activity.
Many hands make for light work.

Torchbearer’s turn system is a vital part of the game. It’s how you keep track of how often the party must consume supplies and how often they gain conditions from wandering through deep caves with medieval technology. Turns give Torchbearer its tactical edge, differentiating it from most other RPGs.

The turn system is also the biggest obstacle to enjoying Torchbearer. By default, the turn counter advances every time a PC takes an action. Each turn that goes by uses up torches and food.* No player can make a roll without consuming the communal resource of time, which induces severe paralysis in the entire party. No one wants to selfishly move the clock forward without workshopping their action through the entire table. Instead of roleplaying individuals, it feels like being stuck in an analog hivemind.

Perhaps worse, the turn rules create a sort of temporal distortion at the table. It feels like time moves faster with a bigger group, as each character’s action is its own turn. It also breaks suspension of disbelief that two PCs can’t perform different tasks at the same time. There’s no way for PC Farid to pick a lock while PC Emma is reading some scrolls for clues. By the base rules, they have to happen sequentially. But somehow both characters could work on the same thing together via helping dice. At this point, a lot of players become less concerned with the game and more with trying to understand how the flow of time works in this bizarre reality.

To deal with this, I use my most critical house rule of all: each PC gets one action per turn, which they can use to either make their own test or help another character. You advance the turn counter when one of two things happen:

  1. A PC who’s already taken an action tries to go again.
  2. A PC takes an action that’s clearly dependent on a previous action being finished, like the wizard copying a scroll into their spellbook after the cleric has finished translating it.

It’s polite to prompt your players to see if they want to do anything in a turn, but don’t wait if nothing comes to their minds. Move on to the next player who does have an action planned.

This system allows PCs to act on their own initiative and creates a more intuitive flow of time while maintaining the turn limitation that makes Torchbearer work. The tactical element of choosing what to do with your time is still there, but it no longer comes at the cost of individual agency. This house rule works so well that I enshrined it in the rules for my Rising Tide expansion, where I’m proud to say it’s served me well.

You might expect that expanding the number of actions a party gets would unbalance the game, but it doesn’t. In most cases, the major obstacles can’t simply be overcome with more actions: players need to find the right actions. At the same time, Torchbearer’s failure rules are robust enough that giving the PCs more actions is often akin to lending them more rope to hang themselves. On the rare occasion when you need to plan a more difficult adventure to make up for the increased PC agency, it’s well worth the trade.

The main downside to this house rule is it makes Instincts a lot less useful. In the base rules, Instincts allow a PC to take their action without advancing the turn counter, which doesn’t matter as much when there are more actions to go around. To address this, I invest Instincts with the power to ignore conditions. If a PC is acting in accordance with their instinct, they take no penalty from any condition except dead, which tends to be a bit final. With this final house rule, you can keep the game moving without disturbing its mechanical balance.

A beautiful quality of the RPG medium is that each GM can customize a system to fit their table. And thankfully, we have games like Torchbearer to provide fertile soil from which our unique gaming styles might grow. Whether you use anything from this list or not, I wish you success in your endeavors to create and share stories at your table.

Do you like Torchbearer?

Play it on the dark seas with our expansion, Rising Tide. Check it out.

Jump to Comments