Five Tips for Running a Mystery Game

Mystery is one of the most common types of roleplaying game sessions, especially for games that have moved out of the dungeon. They provide both a direction and objective for players: solve the mystery! Mysteries come preloaded with suspense and the subversion of expectations, plus they allow players to feel super smart when they figure everything out. More than that, mystery games provide a great excuse for the GM to show off and explain different aspects of their settings. Nothing works better to get players acquainted with all the social rules of a royal court than having them solve the Prince’s murder.

However, it is important to remember that mystery roleplaying games can be especially tricky. In books or movies, creators have plenty of time to set up their elaborate deception. As a GM, you must do it in real time, with an audience of players who are ready to bust out their smart phones at the first hint of boredom. With that in mind, here are a few tips that will help you avoid the much-dreaded blank looks from your players. Each of them has been tested by yours truly through many hours of trial and painful error.

1. Don’t Roll to Find Clues

Clues are such a basic part of mystery stories that some people end up overlooking them. The hero finds a bloody knife of Japanese antiquity hidden under the mattress, and now they know both how the victim died and that the killer likely has a penchant for Japanese culture. What could be simpler?

The natural inclination of many groups is to roll whatever their system’s “investigation” equivalent is to search for clues. Unfortunately, this is a bad choice more often than not. The reason is simple: what if the roll fails? Without a clue, the game often cannot advance. What if the PCs fail their roll to find the bloody knife, and that was the only way for them to figure out the killer’s identity?

In general, it’s much more interesting to interpret clues than it is to find them. How often in a mystery story does the detective find out there was a clue they missed, hidden at the original crime scene? It isn’t very common. When it does happen, it’s usually because a new piece of information has revealed the importance of something the character had previously dismissed.

The GUMSHOE system from Pelgrane Press takes this idea and bakes it right into the mechanics. Instead of rolling to find clues, the PCs roll to interpret the clues that are available. That way, even if they fail, it keeps the game moving. This is the method I generally recommend for all mystery games, no matter what system is being used.

A final, fringe benefit of not making the players roll to find their clues is that it will reduce overdependence on the “investigation” skill or its local equivalent. Anyone who’s played in a mystery game can tell you how easy it is to devolve into constantly rolling the same skill over and over again.

As with any rule there are exceptions, and it can be acceptable to roll for finding a clue when and if not finding it will lead the PCs on an incorrect but still interesting path. Perhaps they already have a suspect, and not finding the bloody knife means they will continue to believe that suspect is responsible for the murder.

2. Err on the Side of Simple

GMs get a lot of their inspiration from the likes of Sir Arthur Conan Doyle or H.P. Lovecraft, and there is a natural desire to create games that have all the complexity of the source material. As you may have guessed, I’m about to tell you not to do that. Authors of prose works have the editing process to make sure their grand idea all comes together in a way that makes sense, while GMs do not. In a lot of ways, a roleplaying session is like the first draft of a story, and on top of that you have to deal with a bunch of PCs who insist on having free will.

That’s why I always advise GMs to keep their mystery nice and simple. The more complex the story gets, the more likely players are to get confused, forget something important, or just plain lose interest. What’s worse, it’s possible for the GM themselves to lose track of what’s happening, and at that point the players are doomed.

The bottom line is that players don’t have the same automatic investment in the story as the GM. They’re playing to have fun, and they aren’t going to enjoy having to slog through an obtuse mess of red herrings and plot twists just to find the brilliant way it all supposedly fits together at the end. When planning a mystery, ask yourself if the players will enjoy the whole thing, not just the ending. If the answer is no at any point, then that’s where you need to do some work.

This doesn’t mean you can never have a mystery so detailed and intricate as to make Dame Agatha Christie proud, just that you need to be careful. It’s entirely possible that the players will latch onto your story and gobble up every crumb of information. Once you’ve tried a few simple mysteries out on a group of players, if they seem really into it, feel free to kick your game up a notch. It’s a great feeling to know that the players are so invested in your grand story, and the best way to get there is with a trail of smaller tales to whet their appetites.

3. Remember, Your Players Are Not Detectives

If I had a nickel for every time I sat around the game table, scratching my head at some indecipherable clue while the GM looked on, wondering why we hadn’t figured it out yet, I’d have enough for a candy bar at least. To tell the truth, I’ve been on the GM’s side of that too, wondering why my players weren’t connecting the dots I’d so helpfully laid out for them.

When an author wants the hero to figure out something, they simply write that it happened. So long as it’s believable, then everyone goes home happy. Actual human beings are not nearly so predictable. It’s more than likely that a group of very smart players will get completely stumped, even with all the clues they need right in front of them. There’s no guarantee they will be able to figure out what the GM’s logic was when designing the mystery, and frustrated players quickly turn into bored players.

Very few people who sit down at roleplaying games have any real training in solving mysteries, even if they’re playing characters who do. While the GM should leave some parts for the players themselves to figure out, the heavy lifting often needs to be done by the characters in the form of die rolls. In much the same way you wouldn’t expect the fighter’s player to actually be an expert swordsman, it isn’t fair to expect the detective’s player to be well versed in criminal psychology.

Don’t be afraid to throw the PCs a bone if they get stuck. I’ve known GMs who were so devoted to letting the players figure out the mystery on their own that the whole game dragged to a standstill, and that’s not fun for anyone. When the game isn’t moving, give it a prod. Drop some hints. Reveal new information.

Because players can get annoyed if it’s obviously a hint from the GM’s mouth to their ears, it’s a good idea to have a story element already established that can deliver the needed information. I like to plant NPC foils with shadowy motivations ahead of time for just such a purpose. Because it’s not clear where these NPCs stand, they can shift from silent observers to deliverers of much needed clues without breaking anyone’s suspension of disbelief.

4. Always Have a Twist

I told you earlier to keep the mystery simple, and now I’m going to tell you it needs to be at least a little complicated. Mysteries where the PCs guess the outcome after five minutes tend to be incredibly unsatisfying. Players, as the audience, expect to be surprised on some level by the ending reveal. There needs to be an ‘aha’ moment, and that can’t happen too early or everything afterwards will seem pointless.

Ideally, this twist will be built into the story to start with. The players assume that the murderer must be Mr. Murderface because he was the last one to hold the bloodstained knife, only to find out at the last minute that he was actually protecting the real killer, Mrs. Murderface.

Authors have it easy in this respect because characters in a prose work will not figure out the mystery until they are written to do so. Players, on the other hand, are wily creatures, and it’s entirely possible they will guess who the real killer is from the very start. If the entire group is convinced without a doubt that Mrs. Murderface is guilty, then there’s very little satisfaction in spending four hours of investigation to confirm something they already know.

That’s why the GM should be ready, if necessary, to add another twist to the story, even if they had not originally planned one. This should only be done after careful consideration. Stop and examine the game so far. Have the players already unraveled the whole thing, and will they be disappointed when that turns out to be all there is? Is there a change you can make that doesn’t destroy the story’s cohesion? Ask for a short break if you need one; your players will understand.

5. Don’t Be Afraid to Change the Story

Sometimes, instead of blankly staring at clues provided to them, players will take the available information and run with it…right past what the GM had in mind and into a completely different story altogether. If players interpret the clues in a way you never intended, examination of their idea might show it has merit and fits the information revealed so far.

When they find the bloody knife, the PCs skip Mr. and Mrs. Murderface altogether, deciding this must be a plot by the Yakuza boss they tangled with last session. The GM scrambles to look through their notes, and realizes this is exactly what Boss Meanguy would do. If this happens to you, there are two options. The first is get the players back on the original track, usually by giving them some information that will paint the intended story more clearly.

However, correcting course like that can be a lot of work, and some players may end up disappointed that their clever deduction didn’t fit the game’s logic. The fact that they were interested enough in the story to come up with this new theory shows a high level of engagement that shouldn’t be wasted if at all possible. Just like with adding a twist, the GM should look at their notes very carefully and see if they can make this new direction work.

If the answer is no, then that’s that. Even the master of an interactive story world must bow to consistency, and you shouldn’t go with an idea that damages the overall story, no matter how fun it sounds in the moment.

On the other hand, roleplaying game stories are flexible things, and if the GM is half way decent, there’s a good chance the player’s crazy idea can be made to fit. Instead of a sad story about a husband protecting his murderous wife, now the game can be about uncovering a secret yakuza plot to get rid of competition. GMs, like all creative types, shouldn’t get so attached to their own work that they are blind to a better idea. I’ve seen mystery games completely derail because the GM insisted on pulling us back to their original story no matter how much the evidence pointed in a different direction.

The most basic key to mystery games is a great deal of awareness by the GM. Most stories involve hiding some information from the audience, but mysteries go a step further because the enjoyment hinges on slowly revealing what’s actually going on at exactly the right moments. Since you don’t have a dozen or so drafts to get it right, the most important bit of advice to take away is staying flexible. Problems will arise, be they from players figuring out the plot too quickly or not quickly enough, and you must always keep tools handy to correct them.

Treat your friends to an evening of dark ritual murder. In a fictional game scenario, of course. Uncover your lost memories and save the day in our stand-alone game, The Voyage.

Read more about



  1. JakeS

    I have found that a good way of thinking about mystery stories is to “sell” the players a new clue when they get stuck at the cost of some penalty sufficiently acute to be memorable but not bad enough to be painful.

    Say your PCs are investigating a bank that had its vault burgled. They find that the wall to the sewer has been blown out from the inside, but instead of concluding that the burglars must have smuggled the boomsticks into the vault covertly, they start hunting for an insider in the bank.

    It’s not a bad guess, but the next day another bank across town is burglarized in the same way. If they then go look at the janitorial service company (also not a bad guess at all, and you should not be afraid to tell them that), then a third bank is hit, one that uses a different janitorial firm.

    At some point, they’ll try going over the customer records, and discover that the explosives were smuggled in a month ago under cover of a deposit.

    If they pick the right option in the first go, they get rewarded for being clever, by saving ten banks, and cutting the antagonist off from resources he needs to complete his evil plan. If they persist in chasing red herrings, they will eventually accumulate enough data points to solve the problem, and will then have gone through a whole tension-building arc to a dramatic resolution.

    Bonus points for ending the escalation chain with the (now greatly buffed) antagonist directly confronting the characters.

    • Oren Ashkenazi

      That’s a good technique, so long as you’ve got that kind of cost to exact. It works great in an example like yours, though perhaps not so well if it’s a murderer killing people the PCs care about. Then they might feel like they’re being punished for not guessing correctly.

      I think for that kind of situation, you also want to make sure not to spend too much player time on such incorrect leads. They say “We think it might be an inside job,” you say “Well, let’s see what your investigation turns up…” then you time lapse to when the next clue pops up. That way the players won’t feel like they’ve spent a lot of time on something that didn’t matter.

      • JakeS

        The trick I normally use is to make the tangents meaningful, rather than cut them short. But that may be because I tend toward running intrigue/investigation games rather than straight investigation.

        The typical investigation in one of our games delves into the different intrigue factions, which means that the time is in any case not wasted: What the PCs learn while chasing red herrings during the investigation, they can use to inform their choice of alliances in the intrigue. (And if they score a hole-in-one, you can re-skin the red herrings as intrigue subplots – waste not, want not.)

        I imagine it would not work so well in a pure-form investigation game. And it does require all the intrigue factions to have reasonably developed goals and credible motivations, because the investigation segments let the PCs scratch the paint a little deeper than you normally would.

        It also probably helps that our group’s playstyle tends toward Wide Open Sandbox, and the GM’s two main jobs are to keep the world credible and consistent and to ensure that the PCs are poked whenever they start threading water.

        And it should go without saying, but I’m going to say it anyway: If you make the PCs pay for clues with their NPCs, political capital, or things they have actually spent XP on, your group needs to be extremely clearly aligned on where the character stops and the game world begins, and to what extent the GM is allowed to encroach on the former.

        Different groups will have different conventions for this – our group tends toward the convention that “the player spent XP or other (non-trivial) resources on it” is the demarcation line between GM sovereignty and player sovereignty. However, in the case of NPCs bought with Ally dots, or similarly “intangible” stats, it gets a little more fluffy – typically the GM can take those away if there is a compelling story need, but the XP gets refunded. But while that works well for our group, your group will have its own preferences.

  2. Adam Reynolds

    There are two key ideas I have for mysteries that are not mentioned here. The first is that I would recommend that the PCs brainstorm at several points in the investigation after gathering evidence. The TV series Castle largely featured this idea with the two leads, in which their differing approaches synced nicely. It would work especially well as an opportunity for straightforward roleplaying without mechanics and allow characterization to shine through.

    Regardless of roleplaying, the key to brainstorming is that any idea that the players have should be out on the table such that the GM and all of the players are on the same page in terms of where the investigation should go. It would potentially allow the GM to design the evidence to lead to the intended path, if the players are headed in a different direction than intended.

    A second idea is that instead of just discovering who the killer was, the PCs should have to prove it. This is a frequent concept in mysteries in which the characters know who did it but have no real evidence. This is where I disagree with the idea that it is wrong to make players roll to discover evidence. Failing a roll would not mean that the players don’t find the evidence, but it would mean that they cannot use it to prove the person’s guilt. It could either mean it is somehow contaminated or that their investigation has tipped their hand to the killer. But it would still tell the players who the killer was.

    In terms of mechanics for this, something like Fate’s abstract damage system could work nicely, In that system, each piece of evidence could be an invocable aspect in which the investigators gather evidence in preparation to take out the killer and prove his guilt. The final attack would likely be the final scene in the interrogation room.

    • SunlessNick

      This is where I disagree with the idea that it is wrong to make players roll to discover evidence. Failing a roll would not mean that the players don’t find the evidence, but it would mean that they cannot use it to prove the person’s guilt.

      This relates well to my own attitude to rolling for clues, which is that failure means “clue but with complication.” If it’s a chemistry test on something, you get the clue either way, but a failure means you’ve used up the sample, while success means you have some left to test for something else. If you’re using Call of Cthulhu’s Library Use skill, a failure means getting the information took all day, or needed the help of the librarian who’ll then remember you when cultists ask her about it. That sort of thing.

      One game a ran a couple of years ago had one character who was an archaeology student who fumbled a roll while working on a dig, and decapitated a statuette that turned out to be Orcus (Etruscan Orcus, not D&D Orcus). When the hotel she was in was besieged by hungry ghosts the following night, the poor player spent the whole game convinced it was her fault. Eliminating rolls has a lot to recommend it, but couldn’t have delivered that.

  3. SunlessNick

    On the theme of being willing to change the story, the film Gone (*not* Gone Girl) had an interesting technique, in that it front-loaded with more half a dozen “it’s totally obviously them” suspects – not in terms of evidence, but in terms of being the kind of people who turn out to be villains in films like this – including a scene suggesting the guilty party was the protagonist herself. As well as making it easy to change tack if players latch onto a different one than you intend, it also counters figuring out the miscreant through story dynamics rather than following the evidence.

  4. Kenneth Mackay

    The tip I’d add is – get your players to WRITE THINGS DOWN!

    In my experience, players remember their pet theories about the mystery far more than they remember the clues you’ve given them (especially any clues that don’t fit their theory) and won’t recall the name, or even existence, of the NPC they questioned a couple of weekly sessions ago, much less the things they said that make sense of the clues they’re finding now…

    • Oren Ashkenazi

      One thing I have to look out for is players talking themselves out of clues. “No folks, I promise the bloody knife is just a bloody knife, it’s not secretly food coloring. I’m not that clever.”

Leave a Comment

Please see our comments policy (updated 03/28/20) and our privacy policy for details on how we moderate comments and who receives your information.