How to write a fight scene readers actually feel
The most common fight-scene failure isn't bad choreography — it's choreography at all. A paragraph that tracks every strike, parry, and footwork change reads like a referee's transcript: technically complete, emotionally empty. Readers don't experience a fight as a sequence of moves. They experience it as a character in trouble, perceiving fragments, making split-second choices with something to lose. Write that, and the scene works even if you've never thrown a punch.
Choreography versus experience
A fight on the page is not a fight on a screen. Film gives the audience the whole frame; prose gives them one consciousness under pressure. The blow-by-blow approach fails because it narrates from the referee's seat — objective, complete, and outside anyone's skin. 'He threw a left hook; she blocked it with her forearm and countered with a knee' is information. It isn't experience.
What a person in a fight actually gets is fragments: the wrongness of a shadow moving, an impact they don't feel until later, the taste of copper, the single detail that weirdly sharpens — a torn shirt seam, the smell of rain. Write the fragments your character would get, in the order they'd get them, and let the reader assemble the fight. You need enough spatial logic that the reader is never lost — who's where, what's in reach — but the camera stays behind your character's eyes, not above the ring.
Rhythm is the speed: cut the sentences down
Prose has no frame rate, so speed has to come from syntax. Long, subordinate-clause sentences read as slow no matter what they describe; short declaratives read as fast. In a fight, shorten everything. Fragments work. One-line paragraphs hit harder because white space reads as time compressed.
Cut the connective tissue that survives peaceful scenes: the 'he began to,' the 'she felt herself,' the stage directions between beats. Verbs carry action scenes — pick concrete, violent-specific ones and skip most adverbs, because 'hit him hard' is weaker than 'hammered him.' Then, when the fight turns — a weapon drops, an ally goes down, the character realizes they're losing — let one longer sentence land like a held breath before the rhythm snaps short again. Contrast is the instrument; all-short is as monotone as all-long.
Read the scene aloud at drafting speed. Anywhere your mouth wants to skip ahead of the words, the sentences are too long for the moment they're carrying.
Stakes over blows: why the reader cares
No fight is interesting because of the fighting. It's interesting because of what the fight decides. Before the first blow, the reader should know what losing costs — the gate closes, the witness dies, the secret gets out — and ideally the fight should be about something the combatants can't just punch: time running out, a third party in danger, information one side must not reveal.
This is also the answer to the pacing problem of long fights: escalate the stakes, not the moves. A fight that goes 'punch, bigger punch, biggest punch' flatlines; a fight where the floor starts to burn, then the hostage wakes up, then the hero's weapon breaks keeps rising because the situation deteriorates, not just the bodies. And let your character make decisions mid-fight — sacrifice position to protect someone, choose the risky move because the clock demands it. Decisions under pressure are characterization; trading blows is not.
- Establish the cost of losing before the first blow lands.
- Escalate the situation, not just the violence.
- Give the POV character at least one real decision mid-fight.
- End on the consequence, not the final blow — the fight matters for what it changes.
The injuries have to persist (this is a continuity problem)
Nothing breaks reader trust faster than a hero who takes a knife to the shoulder in chapter 12 and climbs a cliff two-handed in chapter 14. Injuries are promises: the pain, the limp, the field dressing, the weeks of favoring one side — that's where the fight's cost becomes real, and cost is what made the fight matter. If the wound vanishes, retroactively the fight didn't count.
Practically, this is a tracking problem, and it compounds across a series. By Book 3, your protagonist has a history of damage — the scar on her forearm from the Book 1 duel, the knee that never fully healed — and readers with the whole series on their shelf will notice when the scar switches arms. This is exactly the kind of fact worth keeping in structured canon rather than memory: Scribegrove's series-level world model holds character details like injuries and abilities on the series itself, so every book inherits them, and Grove can flag the chapter where the healed hand does something it shouldn't. However you track it, track it — write the injury into the canon the day you write it into the scene.
Frequently asked
How long should a fight scene be?
Shorter than you think — most real fights are seconds long, and most published fight scenes run a few hundred words to a couple of pages. Length should track stakes: a back-alley scuffle gets paragraphs, the duel the whole book aimed at can carry a chapter. If the fight is long, it needs escalating stakes, not more choreography.
Do I need to know martial arts to write fight scenes?
No. You need spatial logic (who is where, what is in reach), plausible cause and effect, and a POV character whose perception you render honestly. Research helps you avoid howlers — what a concussion actually does, how fast someone bleeds — but readers follow the experience, not the technique names. Expertise on the page often reads worse, because it tempts you back into choreography.
This guide is general information for authors, not legal advice. Platform and store policies change — verify the current terms wherever you publish.