Running Large-Scale Combat in D&D

Gold, Green and Red is a war story. Juramentum, a city of four thousand, defends itself against forces that outnumber it in every engagement. Across 88 episodes, our GM Angel Gammel ran siege battles, field engagements, and urban combat with hundreds of combatants on each side. Here is how he made it work at a table with eight players and no miniatures.

The Mind's Eye Over the Grid

Angel does not use battle maps for large-scale combat. This was a deliberate choice from Episode 1. When the Utini assault hit Juramentum's walls, there was no grid, no miniatures, no measured movement. There was Angel's voice describing what the players could see from their position on the wall, and the players' imaginations filling in everything else.

This works because large-scale combat is not about positioning. It is about decisions. The players do not need to know exactly where every enemy unit is. They need to know what is happening in their immediate vicinity and what choices they have to make right now. Angel gives them a sensory snapshot: what they see, what they hear, what they smell. Then he asks what they do. The battle exists in the space between the GM's narration and the player's response, and that space is larger and more vivid than any grid can contain.

The first AfterDarke podcast episode covered this exact topic. The mind's eye, as Angel calls it, allows the GM to shift scale moment to moment. One sentence you are describing the entire battlefield. The next sentence you are inside Merick's head as he parries a blow. The grid forces you to stay at one altitude. The mind's eye lets you fly.

Abstract the Army, Focus the Drama

The Battle of the Broken Tower spanned Episodes 12 and 13. One hundred and thirty-two defenders fell. The tower itself was renamed Yulia's Watch after Granny Yulia Touchgem called down lightning from a clear sky and did not descend. This was a battle involving hundreds of combatants, but the players experienced it as a series of personal moments connected by a rising tide of chaos.

Angel's technique: the army operates on a simple morale and momentum system that he tracks behind the screen. The players never see army hit points or unit positions. Instead, they hear the results. The wall is holding. The gate is under pressure. The north approach has been breached. These reports arrive between the players' turns, delivered by NPCs who are out of breath and terrified and sometimes wrong about what they saw.

The players' actions influence the army's momentum. When Merick rallied the Myrmidons at the gate, the army's morale shifted. When Marcus channeled golden light in the Healing Hall, the wounded returned to the fight faster. The players are not commanding the army. They are the fulcrum points around which the army's fate turns, and that is more dramatically satisfying than any command structure simulation.

Make NPCs Expendable

One hundred and thirty-two defenders died in the Battle of the Broken Tower. That number became the torch vigil. Every new Myrmidon stands watch for one hundred and thirty-two hours before earning their title. The number is specific because the dead were specific. They had names. Some of them the players knew.

This is the hardest part of large-scale combat for most GMs. Named NPCs have to die in front of the players. Not off-screen in a summary. At the table, in the moment, where the players can see the consequences of the battle they are fighting. When Beacon Gaius flung himself before a lethal stab meant for Marcus, that was not a stat block being removed from play. That was a person the players had spoken to, argued with, and relied upon, making a choice that the players could not undo.

The rule Angel follows: in every large battle, at least one named NPC that the players care about does not survive. Not because the GM is being cruel. Because war is not a game mechanic. It is a story about loss, and stories about loss require actual losses.

Give Players Impossible Choices

The best moment in any large-scale battle is not a critical hit. It is a choice between two things the players cannot afford to lose. The wall is cracking and the Healing Hall is on fire. You can save one. Which one? Merick is outnumbered at the gate and Marcus is channeling a spell that will break if he is interrupted. Who do you protect?

Angel designs every major battle around at least one impossible choice. The players cannot be everywhere. They cannot save everything. The tension of large-scale combat comes from scarcity: there are more problems than there are heroes, and the problems are all urgent at the same time. The players do not experience the battle as a series of encounters. They experience it as a series of sacrifices, and each sacrifice reshapes the battlefield in ways they did not expect.

Track Consequences, Not Rounds

After the Battle of the Broken Tower, Juramentum was permanently changed. The Myrmidons were decimated. The Obsidian Guard was formed from freed slaves to fill the gap. The watchtower was renamed. The torch vigil was instituted. One hundred and thirty-two torches burn every night, and the vigil has not been interrupted since.

Angel does not reset the board after a battle. The damage is permanent. The dead stay dead. The buildings that burned stay burned until someone rebuilds them, and rebuilding takes sessions, not spell slots. The players return to a city that bears the scars of what they just fought through, and those scars are visible in every subsequent episode.

This is what separates dark fantasy combat from standard D&D combat. In standard D&D, a battle is an encounter that the party survives and moves past. In dark fantasy, a battle is an event that reshapes the world, and the reshaping is the story.

The Practical Framework

If you want to run large-scale combat in your own campaign, here is the framework that 88 episodes validated.

Before the battle: decide what the army's morale threshold is and what the players can do to move it. Identify two or three named NPCs who will not survive. Design one impossible choice that splits the party's attention. Write three sensory snapshots for different points in the battle (early momentum, crisis point, resolution).

During the battle: narrate the army as weather. It surges, it breaks, it reforms. Between player turns, deliver battlefield reports through NPCs. Let player actions shift momentum visibly. When an NPC dies, let the moment breathe. Do not rush to the next initiative count.

After the battle: do not skip the aftermath. Play the dawn. Let the players walk the field. Count the dead. Name them. The aftermath of a battle generates more character development than the battle itself, and it costs nothing to run.