I came across a video (helmet camera) of an Taliban force ambushing US infantry troops in Afghanistan. This is a link to the video of ambush: Combat in the First Person, Spc.Michael Gannon guides us through a mission in Haruti by sharing the video he captured on a helmet camera. -- note that there is swearing but no gore/death therein. The ambush and resulting fire fight goes on for six hours. This is clearly a realistic situation. Were someone run a modern day game, this would not be an out of the ordinary encounter for either soldier or insurgent characters.

What RPG combat system would model this situation?

How would you run this as an encounter within the recommended system?


9 Answers 9


I enjoy (not all the time, but some of the time) games that try to very strongly simulate the real world and have asked the same question, as I do enjoy such a simulation. So you're not alone, don't let the 4e/indie crowd convince you "no one wants that." They don't and that's fine, but if you do, read on.

The pace of RL combat is mainly affected by the Clausewitzian principles of:

  1. Friction - in gaming terms, Initiative
  2. Fog of War - in gaming terms, Perception

Friction is basically the mass of uncertainty that keeps most folks huddled under cover instead of running around like Quake deathmatchers. It's partially fear of death but a host of other factors that make the RPG paradigm of "doing something every 6 seconds without fail" patently unrealistic. Some games have an initiative system that somewhat simulates part of this effect; for example in Alternity you can lose one or both of your actions in a round if you roll poorly enough. You can also add bits of uncertainty to the rules to simulate friction - like one problem with many games is that movement rate, unlike every other part of a character, is fixed not variable, allowing confident exact maneuver. I often use a "Move check" and if you bork it you just may be exposed an extra round while trying to cross that alley under fire...

Fog of War is basically you not really knowing what is going on (this feeds into friction). In most RPGs you see all opponents all the time; making some kind of Perception check to detect them is by far the exception. But even some time outside playing paintball reveals to us that we are woefully unaware of our surroundings, where the enemy is, and where our allies are at any given time. Ironically one of the best ways of simulating this might be to incorporate computers into gaming, such that only opponents (and allies and terrain features) someone sees themselves would be revealed to them. You'd get a hugely different combat dynamic! Much of the firing in RL combat is basically blind and has less than even the usual minimum level of RPG granularity ("1 in 20") chance of hitting.

Both of these factors conspire to get you the slower move to decision that distinguishes real life combat from RPG combat, despite real combat often being "one hit and you're out of the fight." (Ablative hit point models are an attempt to approach this problem indirectly - by reducing one-shotting they try to get the longer battle times.)

Not many published RPGs deal with this topic successfully. Palladium Games' Recon!, a Vietnam war RPG, made a faithful effort towards it - an example is ambushes specifically, the ambushers vs the ambushees get very large bonuses/penalties to what they are doing to reflect the confluence of these factors. I've played it and this game does a good job of simulating the uncertainty of combat.

[Edit: I was in Half Price Books this weekend and was reading through Blood & Guts, a modern warfare supplement for d20 Modern, and it had distinct sub-rules for those wanting a more realistic combat experience.]

An ex-Marine buddy of mine wrote a couple really good articles (oddly enough, for "The Way, The Truth, & The Dice," the e-zine of the Christian Gamers Guild) about adding real world data-based realism to combat, using GURPS as the proximate rule system. The first article covers hit location/effects and the second covers friction (see page 22 of the PDF). They are worth mining for ideas.

  • 6
    \$\begingroup\$ "You don't see any targets, but there are bullets whizzing over your head from every direction. Your heart is hammering in your ears, and you're certain you're going to die. What do you do?" Narrate the non-attacking decision points, presenting each turn of events according to their choices, including the occasional "OH SHIT THERE'S THE ENEMY" opportunities to attack. \$\endgroup\$ Aug 19, 2011 at 18:13


I've said it before - I know this answer gets boring. But if you want a system that starts with baseline reality, is your answer.

Now, I'm not talking about modeling the wait. I think that would be less than fun at the table. I would describe the waiting and the tension in a couple of sentences. Maybe a few rolls for positioning, and the cat-and-mouse game of stealth vs. perception. And then run the few seconds of actual combat.

It sounds like what you're looking for is a game where firearms are dangerous, even to PCs. Where someone will consider each move because any hit could be a deadly one. Well, that's GURPS.

Let's look at an "average" case:

  • An average person in GURPS has 10 HP.
  • A typical assualt rifle round does 3d6+2
  • 3d6+2 rolls an average of 12.5
  • My memory is fuzzy here, but I seem to remember that bullets do piercing damage
  • Injury from piercing damage is 1.5x whatever penetrates armor
  • 12.5 * 1.5 = 18.75 - so let's say an averge injury from a single bullet is 18 points
  • That's more than enough to put someone down, causing rolls for unconscousness and even death. I seem to recall that GURPS death is at negative HP

So in one, "average" shot, a regular person is 2 HP from death's door and deteriorating. That makes people consider their actions. It makes them make use of cover. It makes them crawl to stay low.

GURPS has the following addional benefits for a game like that:

  • It can be run with as much or as little detail as you'd like. Maps and minis? Sure! Abstract narrative combat? Fine!
  • It has a tremendous depth and breadth of modern equipment available. See GURPS: High Tech and GURPS: Loadouts books
  • The core system is simple and rational. You can learn it and run it without devoting a year of your life to learning it.
  • It's pretty quick, too: Declare maneuver. Roll to hit if there was an attack. Roll active defense, if applicable. Roll damage if there was a hit. Next.
  • You can go from here to whatever you want. Are your soldiers about to uncover a nest of vampires? GURPS has you covered. Are they about to fall through a hole in time? GURPS has you covered. Are they about to make first contact with aliens using Earth as a hunting ground? GURPS has you covered. And if it never goes any of those places, it's got enough meat to power whatever game you have planned.
  • 6
    \$\begingroup\$ That's not, however, what happens in an abush, Gomad... it's often several minutes of repeated wait, move, popup, shoot, drop, and move on. GURPS can't even replicate a paintball match well, let alone the realities of 6 hours of engaged combat, as the time scale is 1 second. There's even mention of this issue in at least one GURPS 3E supplement. \$\endgroup\$
    – aramis
    Aug 19, 2011 at 20:21
  • 1
    \$\begingroup\$ @aramis - So run it in iterative cycles: A few rolls to determine the outcome of wait and move; a couple of seconds exchanging fire; repeat. Do you really want to model each moment of hiding and positioning? Or do you want to know the result of each extended period, in terms of advantage / disadvantage for the next few seconds of fire and cover? \$\endgroup\$
    – gomad
    Aug 19, 2011 at 22:02
  • \$\begingroup\$ I run PD1E... yes, I do want those enforced down times! Seriously, the combination of 1 second rounds, and acting every round without fail, makes GURPS just about the worst replication of real fire combat I've seen. (It's actually pretty good at replicating SCA duels... see Roleplayer #1 for why.) People just don't react that consistently that quickly. It's CORPS' big breakdown, too. \$\endgroup\$
    – aramis
    Aug 20, 2011 at 6:51
  • 1
    \$\begingroup\$ I'd recommend GURPS Mass Combat for these sorts of large battles without modeling every single projectile and action. That said, if you want to model every single projectile and action, perhaps a computer-based simulation would be more to your liking. ;) \$\endgroup\$
    – user1147
    Aug 23, 2011 at 19:48

Prime Directive 1st Edition

The system has a single roll resulting in both a level of action and an initiative total. It is not uncommon for "joe normal" type characters to wind up not acting other than to dodge for extended periods, and even competent soldiery often have only a minimal level of action.

In short, it results in a lot of "I do nothing this round." While it's actually pretty realistic, and replicates TNG phaser ground battles really well, it's not nearly as much fun as a less realistic but higher level of consistent action combat result.

The system uses a TN set of 4/6/8 normally; A normal man has 3d6 for dice, and a single die has to beat the TN, noting that a 6 open ends, totalling 5+1d6. This means 1/8 of the time (12.5%), one doesn't get an action at all. 98/216 (45%) times, one gets a minimal level (1/2 move or shoot). And only 919/5832 (15%) of rolls get a complete (Move and shoot, or stand and shoot twice). (the rest of the time, it's a full move, or a half-move + shoot.) Note that Armor drastically decreases the chances for higher levels of action...

Still, PD is unrealistically quick, in that rounds are 6 seconds long, and most competent marine PC's will be acting every round, and that 6-hour battle won't happen, even using the full extent of the rules and firearms...

Note that later PD products are not even ports - they share the setting, but are entirely new take on the setting, theoretically done systemless, then having rules matched to specific editions. GPD is purely GURPS Lite SFU-Trek; PD20 is d20 SFU-Trek, and PD20M revises to be based in D20Modern instead of stock D20. Traveller Prime Directive is announced by Mongoose, and will use (more or less) the MgT core rules.


A friend of mine uses the "9 mil test" as a yardstick to judge how realistic a combat system is: If a 9mm pistol fired point blank into the head doesn't do enough damage to kill a PC then the system isn't realistic.

First edition Conspiracy X passes this test and is our favourite for modern combat situations - players try and avoid gunfights wherever possible as they are deadly or require hospitalisation to recover from serious wounds.

Having said that, if I were running an ambush scenario in isolation I would use a wargame rule set - I've not played it but, from what I've heard, Force on Force from Ambush Alley looks like the best for modern asymmetrical warfare. It would be easier to adapt this down to a roleplaying style encounter than it would to adapt any RPG system I've played up to the scenario you mention.

  • \$\begingroup\$ +1 that's not too dissimilar to the test I use. The second is how a Muay Thai boxing match of 5 three minutes round would take to play and how likely that a critical will be scored in that time. At one second per round, that's 900 rounds. At 30 second per dice roll/resolution, that's 900 minutes of play assuming there is only one roll per person fighting... yawn \$\endgroup\$ Aug 22, 2011 at 15:37
  • 1
    \$\begingroup\$ @gomad: 30 second per dice roll is not unrealistic. I did some statistical analysis (highly academic, with proper timing, and many hours of R modelling... or something) and it seems to be in the bulk part. If a system has to hit, to dodge/parry, to damage, to soak rolls, that's about 2 minutes per exchange. Rinse, repeat. Of course, it depends on how many dice you roll, how much algebra you need to do, and how many tables you need to consult. \$\endgroup\$ Aug 22, 2011 at 15:48
  • 2
    \$\begingroup\$ That's one reason my preferred realistic combat system includes outcomes where the combatants break and circle, which is often what takes up a lot of time in a well-matched fight. \$\endgroup\$ Aug 22, 2011 at 16:48
  • \$\begingroup\$ For all of these reasons a look at 'A Time of War' might be warranted. \$\endgroup\$ Aug 25, 2011 at 1:00

The out-of-print Twilight: 2013 handles modern combat remarkably well, with an excellent focus on the psychological aspects of it.

There is even an underappreciated mechanic of push vs hold every turn: If both sides push, combat continues. If only one side pushes, they gain surprise/advantage. If both sides hold, then combat hits a lull. The pace then slows as sides jockey for position, perform first aid, try to simply spot the enemy etc.

So it would seem the side that pushes will win vs the side that fails to even once. Except that you can't always push, especially with combat stress factors (getting shot at with automatic weapons is more stressful, so is explosives used against you, and fire, etc). You can't do much more than slow down the pace of bleeding out when you're pushing (and it's a lethal game).

Then there's the longer-term mechanics of combat stress and psychological damage...

Crying shame that the book is laid out the way you'd tell someone about the game, vs the way you'd teach someone the game. That and an entrenched franchise fanbase not keen on changes doomed it. Still, pick it up for some good ideas.


Silhouette/SilCore mechanics have been designed to slide easily from personal to vehicle/unit scale, and are pretty quick to operate.

(Most of their games have a wargame part and an RPG part, and the rules are designed to allow the two to be mixed).

So maybe you could resolve most of the 6 hours battle as a squad-level simulation and "zoom in" to the individual (RPG) level whenever the PCs squad is involved in an actual fighing engagement.


TL;DR: Don't just look at system, but your approach to a protracted battle. Find a system with realistic damage ratios, and simulate most of the battle through skill checks with possibly dire results for failure.

The big problem

I understand that you are asking for a game recommendation, and I will list a few, but the problem isn't just the system you might use but also how you approach a battle as a whole.

Most systems use a short duration combat turn that is between 1 to 60 seconds, typically sitting at 6 or 10 seconds per 'round/turn'. Almost any system, due to this, will need to greatly abstract the 'standard combat system' for the general gunfire and cover/shoot/move aspects by warping time as you go.


Ambushes are VERY deadly in reality. You have someone that is likely completely blind of what's going on being assaulted by an individual, a group, or simply the equipment of people that are expecting absolute control of the situation.

Tactics in modern warfare for an ambush commonly either employ people in well chosen locations firing at a central location or remote explosion detonation. Using both is frequent. If the actual catalyzing event of the ambush is a detonation and it is not set to automatically fire, then the person with the control of the detonation has the ability to choose exactly which vehicle to kill people in.

In reality this generates enough fear and paranoia that even a teddy bear sitting in the road is suspect, as is discoloration of the road or recent repairs.

If you are not the targeted vehicle/person in the ambush, then it can and will likely turn into a protracted battle that can last a long time. This means a lot of bullets flying, but very little movement, movement that is highly controlled, and a lot of cover fire to cause the enemy to pause long enough that you can hope that you won't get shot directly at while moving. This is more of an extended skill check series that has dire consequences for failing than actual combat rounds.

Possible Systems

Almost any system can make this a worthwhile roleplay, just as almost any system can make this a very poor roleplay that takes forever to resolve a single encounter (due to the length of the combat turn). Here are a few systems I might try this type of combat in:

As mentioned by gomad, GURPS does a good job simulating a much more realistic world by keeping the base game very realistic stat-wise and having very realistic applications of armor and cover (making it near impossible to get hit if you protect yourself properly). This is actually my preferred choice, since it requires the least doctoring to get running right.

One of the d20 titles (*d20 Modern** or Star Wars, I think, maybe both) had two abstractions for damage taken. You had your actual health (based directly on constitution), and the traditional hit points, which were listed as an abstraction of near misses, dodging out of the way just in time, exhaustion, and slightly injuring yourself in an effort to get away from the fight. I don't think this is the right fight, but it could work for some types of games similar to this. This is my least favorite system listed here. This is the most likely system to get a character hit by a shot on the way to the shooter though not necessarily the most lethal.

World of Darkness (not the supernatural splat books) simulates a low hit point system, and since by default guns are unsoakable damage it is lethal for normal humans. You don't have to use any of the supernatural stuff associated with WoD in order to create a realistic battle, but should you want to add an element of the supernatural, it's easy to throw in.

Shadowrun is another possible choice. Ambushes do turn ugly very quickly in this game. It's based on a futuristic version of earth, so if you want cyberpunk, then it works well. You can play with or without magic as you feel comfortable with. Characters are a little less squishy than in WoD or GURPS, but that doesn't stop bullets from killing people quickly. You can also declare the guns listed to be equivalent to modern day weapons instead, and get rid of both aspects.

Most systems that try to implement realistic hit points can work, especially if rules for cover are well established.

Doctoring the System/Encounter

No matter what you do, it's unlikely that you want to have a 6 hour fight broken in to even 1 minute combat rounds, since that is going to take you more than 6 hours of real life to complete a single encounter scenario. I suggest modifying how you approach combat completely and breaking away from the 'combat round' to simulate the entire encounter.

D&D 4e had a very interesting idea that had a very poor implementation, but can be used for inspiration. Skill Encounters. An encounter that is completely based around skills. Although they had refined this process by the end of 4th edition's run, it wasn't very popular because it felt rather limiting, but even deadly encounters could be turned in to skill encounters. Although it's fantasy, one of the examples of this was running from a horde of zombies. Failures, especially very low failures, could cause damage or even death, but every single round wasn't covered in the 'combat round' sense. Some skill encounters even resulted in combat on failure of a roll or the encounter itself. The objective was to allow the players to roll for skill checks as they describe approaching the scenario. They have to hit a specific number of successes on these checks before a specific number of failures (and often they needed as many as 50% more to double the number of successes as failures).

How would someone implement a skill encounter for this type of scenario?

Make a difference between a 'combat round' and an 'assault round' to allow for that abstraction and it becomes relatively easy. The initial encounter of the ambush would be a few traditional rounds of combat until the players can get out of the obvious danger. After this, switch to 'assault' rounds (or some other appropriate adjective), making each round a matter of the players positioning themselves.

Note, this is simply one approach that I believe best simulates the experience without turning combat into a slog of misses for 12 hours straight.

Sample Skill Encounter

Since I extracted the idea from D&D, I will use it's basic stat system in my example: (Strength, Dexterity, Constitution, Intelligence, Wisdom, Charisma, plus Reflex, Fortitude, and Will 'saves')

Design the encounter with a host of 'normal skill checks' to use in the encounter. These checks are going to be based on the most common courses of actions:

Move: This should be a check based on either wisdom (perception) or dexterity (stealth). It simulates either timing your movement for a 'safer' period of time, or moving low to the ground and avoiding direct sight for extended periods. If your system incorporates speed as a roll-able stat this would be appropriate as well. This should generate a 'success' for each successful role, while failure means the character is likely shot at during the move (with normal penalties to hit based on circumstance). Success allows them to get a specific distance. This should be abstracted from their movement rate, extrapolated for multiple rounds. 100 meters, 500 meters, or some other appropriate distance for how long you want the encounter to last. (Don't simply use it as a single movement action, or you won't have sped up the combat.)

Cover Fire: This should be some form of mental check to determine timing (int/wis) or ability to force your opponent to stop firing (at least toward your ally) (cha). This would not generate successes for itself, but instead would give allies the chance to move more easily (a significant bonus to ally's checks for the round). ammunition is expended, but cover is provided on a success.

Locate Attacker: Although this seems like an obvious one, don't take it for granted. The attacker will have an advantage if the ambushed characters don't know the direction to move, or don't know the location they are firing from. I would likely require two successes for this skill check. The first to locate the general direction the enemy is in, and the second to isolate the enemy to a specific area. Obviously the first check would be required before the characters can appropriately hide or even move in the general direction of the enemy. It should be average difficulty unless the attack came from a different direction than the enemy (remote detonating devices, for example) or there are multiple directions of fire, and should certainly be easier than locating a specific position. This is a wisdom (perception) check, and obviously, the further the enemy is from their position, the harder it will be to locate them.

Take Cover: Either total cover or concealment is likely preferred, as anything less than that is likely to get shots taken at the character. If concealment, give a dex (stealth) check, if cover, just count it as a success making them impossible to hit. (Concealment only protects you from being seen, cover has physical substance that can stop a bullet.) This should be required of all characters any round they take an action other than staying in the cover/concealment they are already in. It shouldn't count toward successes, but should certainly count toward failures (and the possibility of getting shot) if a role is made. Do not allow players to take a shot at the ambushers until that character knows exactly where the enemy is.

So the primary courses of action are covered, 'Move', 'Cover Fire', 'Locate', and 'Take Cover'. The players will likely come up with other scenarios. Make judgement calls on what effects those rolls will have. Throwing a smoke grenade could provide total cover to a certain distance of movement, thus giving a high advantage of some sort to movement.

Next we need to look at consequences for failure. In this situation the consequences for failure on most rolls will be to be shot at. Apply distance and cover modifiers to make an appropriate value, and let the enemy take a shot at the character. Any other costs of failure can be brought up on the fly. Overall failure of the challenge would be determined by losing people (possibly all of them). To make it more militaristic, you can have a threshold of failures (or casualties) before the order to withdraw is made.

Consequences of success: If the group succeeds at the skill challenge, they can move into position to attack the ambusher. This would move the game back to a few quick combat rounds to resolve the end of the encounter. Every counted success can move the entire party forward the appropriate distance, or can be split so players have to take turns making cover fire/movement actions depending on the desired amount of realism.

Wrapping it up

This type of an encounter system allows the tedious aspects to still provide appropriate levels of danger, while not requiring a roll every 1-10 seconds of game play for every player. Technically it isn't really combat until they are directly engaged with each other, and this is a nice abstraction that allows the battle to move swiftly in RL terms compared to playing the whole thing out as a standard combat.

If you have a system that deals with damage from guns relatively realistically, then you can model this type of encounter system to create a gripping encounter that will definitely be exciting enough to create the atmosphere of a realistic battle, while not being remembered for mind-numbing, time-consuming rolls for hours on end.


I have had good success with older versions of Shadowrun (SR2, SR3) for this sort of thing. Cover makes a huge difference to target numbers (and you can increase the penalties if you want), and you can run a lot of it at long range for the weapons, which makes actually hitting anything a minor miracle unless someone does something stupid.

If you don't have any magical healing and you don't have any advanced tech, you basically don't heal while on the fight, so you'd better be careful. But chances are that if you're careful, you'll receive an injury 1-3 times before dying rather than just getting unlucky and being gone, which generally makes for more fun gameplay. (Whether it's more realistic depends on the scenario.)

But as a more meta answer: maybe you should use any system that you know really, really well. When you know a system very well, it's much more obvious how to bend it such that it delivers the results you want. (Even better if your players also know, and will act accordingly.)


Phoenix Command was developed with the express goal of being as realistic in combat as possible. It also features a setting with roughly modern technology, though it's a bit outdated with the more modern kinds of digital warfare. It's horribly difficult to run, but it does the job of delivering an extremely realistic modern combat experience better than probably any other RPG ever made.
For best results:

  1. (required) Ensure your GM is a tactical genius who has memorized the rulebook.
  2. (optional) Expect to die.

The encounter described is incredibly typical for a Phoenix Command encounter, so I will neglect a detailed run down of the combat here. Basically the ambushees' greater skill and various minor tactical failures on the part of the ambushers leads to the breakdown of the planned short-term ambush scenario in such a manner that a prolonged engagement occurs.


Not the answer you're looking for? Browse other questions tagged .