How to write fight scenes

Fight scenes are the set pieces of action-driven fiction, and in progression fantasy and LitRPG, genres this catalog serves heavily, they are structural: the fight is where a chapter's promises are kept or broken. Fight craft has no single canonical reference, but the working advice converges tightly. Carla Hoch, who writes Writer's Digest's FightWrite column, names the two failures she flags most often in submitted scenes: "too little focus on the sensory experience and defying body mechanics". Her instruction is just as compact: "Put me in the scene of the fight. Make me root for one of them."

Whose fight is it

A fight is experienced by one body. Point of view decides what the reader knows during it: the flanking enemy the viewpoint character has not seen, which blows landed and which only seemed to. Reedsy's guide keeps the interior channel open: brief, instinctive thoughts carry the fight's emotion, and they work "even in third person"; long inner monologues pull the reader out of the action. The language belongs to the fighter too: Jeff Bens, writing at LitHub, says a fight's "word choice, cadence and rhythm" arises from "the emotional world of the protagonist(s)".

That makes a fight the worst place to lose viewpoint discipline. Decide whose fight it is before the first blow and hold one viewpoint per scene (the arguments are in choosing point of view); combat reactions are also where head-hopping slips cluster. In a battle with many combatants, the viewpoint is your depth of field: everything outside it becomes threat, noise and guesswork, usually an asset.

Pacing and sentence rhythm

The folk rule says short sentences make fights fast, and the sources partly agree. Reedsy's advice covers length as well as rhythm: "Most fights only last a couple of minutes", so a single fight deserves a page or two, not a chapter, and a quick, violent fight wants "punchy verbs and brief descriptions".

The stronger reading is that rhythm is a dial, not a rule. Bens: "A fight scene might use short sentences or long, quick or slow." A punch, he notes, is "a direct physical experience most likely requiring direct language". At the climax the scene "may in fact slow down, with the character's sense of time expanding". The consensus: track the tempo of the exchange and the perception of the viewpoint character. Staccato for the exchange of blows; longer sentences where the character's attention dilates.

Choreography the reader can follow

• Choreograph for yourself, translate for the reader. Work out every position and exchange, then render it in plain, forceful language. Reedsy warns that clunky technical phrases "will only confuse readers, slowing them down and stealing that sense of adrenaline".
• Act it out. Hoch's method: "you don't have to be able to do the move, but turn your body as the character would". Impossible body mechanics stay invisible on the page until your own shoulders refuse the position.
• Prefer sensation to technique. "Not everyone can relate to a fight, but they can relate to a sensory experience": breath, impact, exhaustion. Her examples are a slumped back, dragging feet, air "exploding from his chest" when a body hits the ground.
• Ground the space. Craft consensus: give the reader two or three landmarks and keep every fighter's position relative to them. A fight fails the moment the reader cannot say where anyone is standing.

Stakes and cost

A fight needs a believable reason. Reedsy names three that can carry one: survival, protection of a person or an ideal, and honor. It also supplies the exchange-rate test: if a character is risking their life, "is their honor worth it?" Underneath sit the ordinary story forces, a want versus need collision with another character's want. Bens locates the engine deeper still: "Characters fight from these wounds."

Cost is what makes a fight an event rather than a display. Reedsy's final step is aftermath: injuries and recovery, shown so the consequences land. A character carved up in chapter four and flawless in chapter five teaches the reader that violence is free; injury and state continuity is the discipline that keeps the bill visible.

The purpose test

The craft consensus, anchored in all three sources: a fight must change something. Bens: "The climax of a fight needs to affect the reader beyond the moment itself", and a fight is not resolved "until the inner world of the character is felt by the reader to have shifted". Reedsy requires every fight to be resolved for the story to move forward. The test is deletion: remove the fight and ask what changed. If nothing did, no plot fact, no relationship, no knowledge, no self-concept, the scene is choreography for its own sake and should be cut or compressed to a paragraph.

Power creep in progression fiction

Progression fantasy and LitRPG make fights the genre's proof of progress, which creates a specific failure mode: power creep. The genre's craft consensus: every escalation needs a cost the reader can track. If each fight is won by a newly unlocked ability, stakes deflate, because the reader learns the outcome is guaranteed and only the flavor varies. The systemic half is power systems, costs and limits; the genre's rhythms are covered in the progression fantasy guide. The fight-scene half: the strongest fights are won at a durable price, a broken tool, a burned bridge, a revealed technique enemies will now counter, so the next fight stays uncertain.

Training and sparring scenes

Craft consensus: training scenes are real scenes with a different job. Their stakes are evaluation, not survival: proving competence, teaching the reader the system's rules, dramatizing the relationship between teacher and student. Hoch's critique of a submitted training scene shows their logic: the teacher punishes a bad habit because "you forget yells faster than you do pain". A training scene works best when it plants what a real fight will later break: the drilled technique that fails under pressure, or the one that saves at cost. The failure mode is comfort, a training arc that runs so long and so safely that the first real fight reads as just another drill.

Before drafting the next one, five questions. Whose head is it in? What does each side want? What changes if the fight is deleted? What does it cost, and is the cost still visible three chapters later? Can a reader sketch where everyone is standing? A fight that passes all five does what the sources ask: it puts the reader in the scene and gives them someone to root for.