Building a Fantasy City: Why It Exists Before What It Looks Like

Most fantasy cities die on the page because they were built as scenery.

The writer starts with the mood board. Gothic spires, floating lanterns, a market that smells of cinnamon and rain. Then they staple a river to it because rivers look nice, and drop a castle on a hill because castles go on hills.

The result reads like a theme park. Beautiful, empty, and impossible to believe in for more than a paragraph.

Real cities are not scenery. They are arguments.

A city is a place where a lot of people decided, over centuries, that it was worth living in this exact spot instead of anywhere else. Somebody had to eat. Somebody had to defend the walls.

Somebody had to get grain from the farms to the ovens before it rotted. Every real city is the frozen record of thousands of those decisions, and the ones that survived did so because the answer to one question was good enough to keep answering it: why here?

That is the question this whole guide is built around.

Get that answer right and the map, the districts, the smell of the place, the reason a war would be fought over it, all of that falls out almost for free. Get it wrong and no amount of stained glass will save you.

Start with Why the City Is Here

Before you name a single street, answer this: if I moved this city ten miles in any direction, would it stop making sense? If the answer is no, you have scenery, not a city.

Look at how real cities got their start.

London is where it is because that stretch of the Thames was the lowest point you could reasonably bridge and still be far enough upriver to be safe from sea raids. Bridge plus tidal river plus defensible ground equals a place worth fighting over for two thousand years.

Venice exists because a bunch of refugees needed somewhere an army on horseback could not follow them, so they built on mudflats in a lagoon and turned the water itself into their walls. Chicago is a swamp that happened to sit at the one short portage between the Great Lakes and the Mississippi watershed, which is a boring sentence that explains why a global city grew out of a marsh.

The pattern under all of them is the same. Cities cluster on the seams of geography.

River mouths, natural harbors, fords and bridge points, mountain passes, the last stop before a desert, the first safe port after one. These are the places where goods have to change hands, or slow down, or pay a toll, and wherever value changes hands, people gather to take a cut.

A city is a toll booth that grew a personality.

The fantasy cities that actually work know this. Ankh-Morpork sits on the river Ankh at the last bridge before the sea, which is exactly why it became Discworld's clogged, corrupt, magnificent trade heart.

Camorr, in Scott Lynch's The Lies of Locke Lamora, is a Venice built inside the ruins of an older alien civilization, and its whole criminal economy runs on canals, islands, and chokepoints you can control. New Crobuzon in China Miéville's Perdido Street Station grew where two rivers meet, and the book never lets you forget that the city runs on the industry and filth that a river junction makes possible.

Minas Tirith is a fortress guarding the one approach to Gondor's heartland, so every stone of it is about sightlines and siege. King's Landing sits on a superb natural harbor at the mouth of the Blackwater, which is why it is both rich and constantly under threat by sea.

None of those cities would survive being moved ten miles. That is the test.

Write down the one geographic fact that makes your city inevitable, and let the rest of the worldbuilding argue back from it.

Geography and Climate Decide More Than You Think

Once you know why the city is there, the terrain around it stops being a backdrop and starts making decisions for you.

This is the part where most guides tell you to "consider the geography," which is useless. Geography is not a vibe. It is a set of constraints that force specific tradeoffs, and the tradeoffs are where the good stuff lives.

A city on a river gets trade and fresh water and easy sewage, and in exchange it gets floods, and it gets attacked from upstream, and it has to decide who controls the bridges. A city on a defensible hill gets safety and long sightlines, and in exchange it gets a brutal water problem, because you cannot easily pump water uphill in a pre-industrial world, so every siege becomes a question of cisterns.

A coastal city gets the whole ocean as a trade partner and gets raided by anyone with a boat. A city in a mountain pass controls a chokepoint worth a fortune and freezes half the year and can be starved out by whoever holds the valleys on either side.

Notice that each of those is a bargain, not a feature. The thing that makes the city rich is usually the same thing that makes it vulnerable.

That tension is the engine of plot. Your city's greatest asset should also be the knife at its throat.

Here is a quick reference for what a given setting forces on you.

Climate does the same work on a smaller scale.

A city that freezes builds thick walls, steep roofs, and a culture organized around getting through winter. A tropical city builds for airflow and shade and lives half its life outdoors.

This is not decoration. It changes the architecture, the clothes, the festivals, and the times of year anyone would bother going to war. If you want to go deeper on translating terrain into a drawable map, the worldbuilding maps guide covers how to keep rivers, elevation, and trade routes honest.

Districts Are the City Arguing with Itself

Now you divide the city up.

The lazy version is residential, commercial, industrial, done. The version that works treats districts as the visible scar tissue of history and money.

Cities sort themselves. The tanneries and the slaughterhouses and the dye works go downwind and downstream, because everyone hates the smell and nobody with power wants it near them.

The rich move uphill and upwind, always, in every culture, because they can. The docks breed a rough, transient, polyglot quarter because that is who works the docks.

Old money clusters near the old center where the temples and the palace already stand. New money builds gaudy things just outside the old walls where land is cheaper and the guilds have less grip. The poor fill in the low, wet, dangerous ground nobody else wanted.

So a district is not a zone on a map. It is an answer to who has power and who does not, written into geography.

Camorr does this beautifully. The nobles live in glass towers on the island called Alcegrante while the con artists and cutthroats work the Cauldron and the Snare, and the physical separation of those islands is the class structure made concrete.

I pushed this idea about as far as it goes in my own world The Bottom, where the entire world is one vertical city ten thousand years tall. There, altitude is the class marker. The sky districts get real sunlight while the deep city runs on bioluminescence, and the Administration owns every bit of vertical infrastructure and charges you to breathe better air the higher you live. The whole social order is just the map turned on its side, and where you were born on it decides almost everything.

When you place a district, you are really placing a piece of the city's politics.

A rough guide to what districts signal, so you can pick on purpose:

You do not need many. Three or four districts with real friction between them beat a dozen that are just differently colored blocks. The friction is the point.

Population Is What the City Runs On

A city needs a reason to feed itself, and that reason shapes who lives there.

This is where a lot of fantasy quietly cheats. A big walled city needs an enormous amount of food coming in every single day, which means a big city implies a productive countryside, good roads or rivers to move the grain, and someone with enough authority to keep it flowing.

If your city is a lonely spire in a wasteland, it needs an extremely good in-world reason to be fed at all. Magic can be that reason, but then magic is now doing economic work, which is a much more interesting job for it than shooting fireballs.

Once you know what the city does for a living, its people follow.

A trade port is loud and mixed and full of languages, because trade drags people together. A holy city is full of pilgrims and the industries that fleece them. A fortress city is full of soldiers and the people who supply, feed, arm, and bury soldiers.

The dominant industry decides the dominant class, and the dominant class decides who is looked down on. That hierarchy is the social spine of the place.

If your world has non-human peoples, resist the urge to give each one its own tidy neighborhood like a food court. Real difference is messier and more interesting when groups are tangled together, competing for the same work, resenting each other over old grievances, intermarrying anyway.

The question is not "where do the dwarves live" but "what do the dwarves do here that no one else will, and who hates them for it." That is worldbuilding with teeth. For pushing the wider culture and history out from the city into the whole setting, creating a fantasy world picks up where this leaves off.

Landmarks Should Point Back at the Reason

Landmarks are where writers reach hardest for spectacle and get the least out of it.

A cathedral or a mage tower or a great statue is only memorable if it is telling you something true about the city. Minas Tirith's seven walls are not there to look cool. They are there because the city's entire reason for existing is defense, so its most striking feature is a defensive one, and that consistency is why it feels solid.

Ask what your city is afraid of, or proud of, or built on, and let the biggest structures answer.

A city that grew rich on the slave trade has a monument it is ashamed of and cannot tear down. A city saved from a flood a century ago has a memorial nobody visits until the water rises again. A city ruled by a merchant council has a stock exchange more impressive than any temple, because in that city money is the real god.

The landmark is an argument the city is making about itself. Make sure it agrees with the reason the city is there in the first place.

Adding Magic Without Turning the City into a Toy

Magic is the moment most fantasy cities lose their grip on reality, because writers add it as flavor instead of infrastructure.

If magic exists and it can move water, heal wounds, light streets, or send messages, then it is a public utility, and public utilities restructure everything around who owns them.

The honest question is not "what cool magic is in this city" but "who controls the magic and who gets locked out of it."

If light spells exist, are they cheap enough that the poor quarter has them, or is magical light a marker of wealth that quite literally divides the city into the lit and the dark? If healing magic is real, does everyone get it, or does a temple ration it and turn medicine into leverage?

Magic that follows real rules and real scarcity generates class, resentment, and plot the same way any other resource does. Magic that is just ambient sparkle generates nothing.

This is the same trap as everything else in this guide. A magic system is not set dressing, it is an economy, and an economy has winners and losers.

If you want the mechanics of building one that behaves like a system instead of a special effect, types of magic systems and the broader magic systems hub are the right next stop, and the wider worldbuilding category ties the city back into everything around it.

Bringing the City to Life on the Page

You have done the hard structural thinking.

The last job is to render it so the reader feels it, and here the rule inverts. In the planning you thought like an economist. On the page you write like someone standing in the street.

Do not describe the city. Describe a person moving through one square block of it and let the reader assemble the rest.

The reader does not need the full history of the docks. They need the specific smell of the tide going out, the specific coin a specific beggar is turning over, the specific way the guild banners get taken down before a riot.

Concrete beats comprehensive every time. All that structural work you did is not there to be described. It is there so that every small detail you do put on the page is quietly consistent with a real place, and the reader can feel that consistency even when they cannot name it.

That is the whole trick. Build the city as an argument, then only show the reader the parts of the argument that a character would actually notice.

For turning the setting into scenes and plot rather than a static map, how to write fantasy novels carries it forward, and the broader fantasy writing guide covers where the city fits in the rest of the craft.

FAQ

How big should a fantasy city be? Smaller than you think, if you want it to feel medieval. Most pre-industrial cities held tens of thousands of people, not millions, because feeding and moving a large population without modern farming and transport is genuinely hard.

Ancient Rome hit around a million and it took an empire's worth of grain fleets to keep it fed. A "great" fantasy city of 50,000 to 200,000 is plenty large to feel overwhelming, and it forces you to think honestly about where the food comes from. If you want it bigger, give the world the infrastructure or magic to justify it.

How do you name a fantasy city? Name it the way real places got named, which is rarely on purpose. Real city names describe geography (a ford, a river mouth, a hill), honor a founder or a god, or are worn-down fragments of a much older language nobody speaks anymore.

That last one is gold. A name that clearly meant something once and has since eroded makes a city feel ancient. Avoid names that are just two fantasy syllables plus an apostrophe.

Decide what language named the place and what the name originally meant, then let time sand it down.

How detailed should a fantasy city map be? Detailed enough that you never contradict yourself, and no more. You need to know the real geography, where the walls and gates and main roads are, where the districts sit relative to the water and the high ground, and roughly how long it takes to cross.

You do not need every alley. Readers do not track street layouts, they track relationships (the docks are downhill from the palace, the slums are across the river). Map the logic, not the pavement.

How many districts does a fantasy city need? Three or four with real tension between them beats ten that are just labels. You want enough to represent the city's core conflict, usually a power center, a money center, a working or foreign quarter, and the ground where the ignored people live.

Every district you add should introduce a group with something to gain or lose. If a district does not change how someone in the story could get hurt or rich, it is decoration, and decoration is what makes cities feel fake.

What makes a fantasy city feel real instead of generic? One good answer to "why is this city here and not somewhere else," followed by the discipline to let everything else agree with that answer. Generic cities are collections of cool features with no underlying reason.

Real ones have a geographic or economic logic that shows up in the districts, the population, the landmarks, and the things people fight over. When the reader senses that the city would have consequences if the river dried up or the pass closed, you have built a place instead of a set.