Medieval Demographics in Brief

Historical map of Kudna area 1870s

When I’m developing a wilderness map, I always have to decide, “At what frequency should I place villages, cities, and castles here?” Below you'll find a short document that I've drafted on research for a model on medieval-level demographics. Of course, I'm not the only person to do this, and in fact I don't think this is the only time I've done this; but the document includes copious endnotes so I hopefully stop re-doing this and forgetting which parts are based on real-world data, and which parts I made up for the game.

Here's an executive summary. We use as the basis for our model data from England around the 11th or 12th century;  the population then was around 2 million. (Earlier in the Dark Ages it might have been half or one-quarter this level; later in the High Middle Ages it would be around 5 million, and then collapsing back down to 2 million at the time of the Black Death.) For our baseline in history we find, in round numbers:

  1. Cities: One per 1,000 square miles.
  2. Castles: One per 100 square miles.
  3. Villages: One per 10 square miles. 
(When we say "cities" here we really mean any walled towns.) A variety of examples from Gygaxian rulebooks shows a markedly less dense population, by roughly an order of magnitude (about 10 times fewer features per equivalent land surface area). For game campaign maps, this author is currently using a value about halfway between the two, that is, about one-quarter the baseline medieval population density (i.e., Dark Ages levels). More precisely, we assume that only about one-quarter of the land area is inhabited by humans, with density in those areas approximately the same as medieval England (the rest of the map is left over under the control of various nonhuman races and monsters; in particular, almost any non-plains locations). Therefore this author currently uses, for fantasy realms:
  1. Cities: One per 4,000 square miles.
  2. Castles: One per 400 square miles.
  3. Villages: One per 40 square miles.
This can easily be converted to a recommendation or average on a game map per-hex basis with a little math. Note that the area of a hexagon is \(S^2 \cdot \sqrt{3}/2 \approx S^2 \cdot 0.866\), where \(S\) is the length of the hex measured across opposite edges. (For example, a 24-mile hex is close to 500 square miles; a 6-mile hex is about 30 square miles.) We currently use something like the following map scales:
  1. 100-mile hexes: Shows capital Cities (1 per 30 hexes).
  2. 24-mile hexes: Shows walled Towns (1 per 8 hexes)
  3. 6-mile hexes: Shows baronial Castles (1 per 12 hexes)
  4. 1½-mile hexes: Shows individual Villages (close to castles)
We can otherwise express this by stating that, on average, at Level 1 there is one walled town per hex; at Level 2 one castle per hex; and at Level 3 one or more villages per hex, and therefore these features are not shown on the maps at those scales. The 6-mile hex scale is expected for use in standard wilderness adventure and exploration (smaller sizes permit the entire map to be crossed in a single day, and are generally only used for introductory level adventures, or for added flavor, not strategic movement). We emphasize that at this exploratory level, it makes no sense to depict individual villages (historically, there would be multiple villages per hex); instead, we place castles as shown and assume that each of the adjacent hexes has between 1 and 3 villages in fief (within range of daily patrols, or for a runner/rider to get help within an hour or so). We further assume that the villages mostly have no amenities of interest to PCs (inns, taverns, general stores, etc.) In general we find this makes for attractive and gameable maps, with plenty of space left over for exploration and monster lairs.

Finally, populations for each of the urban features can be simply approximated by the following (this generally hews pretty closely to both historical data and classic D&D rules):
  1. Cities: 10,000-60,000 people.
  2. Towns: 1,000-6,000 people.
  3. Villages: 100-600 people.
A further research question might be this: Is the population scarcity shown in the writings of Gygax consistent with other works of fantasy literature? While likely harder to quantify, this author would hypothesize so. (Consider: Featuring only a single central city in an area that would historically have several, needing to focus the narrative on a few key features, etc.)

See full details in the document below, with appropriate research citations (PDF):


  1. Your choice of 11th and 12th c. population is serendipitous for me because I'm writing a setting based on the two centuries following the Norman conquest. What I found was that the Marquesses on the border of Wales imported a lot of peasants from the continent.

    I wonder how many "a lot" is in terms of population density? Definitely 5% and maybe 10% at minimum.

    On cities: having one city is shorthand. The players know where to go, and everyone important can be gotten to that day (if not actually met with.)

    For the purposes of the game, you can have other cities but they are superfluous. If you want to have a war there can be another city Over There, but it can just be a dot on the map.

    In fact, not having other cities encourages the several players to spread out and make their own key settlements when the time is right.

    Now why would you choose to quarter the researched population density? Is it because your Realm has so much nonplains terrain or is there another reason? How do people administer to such large, sparse realms? Magic or magical beasts?

    1. Cool stuff you're working on!

      Re: "Now why would you choose to quarter the researched population density?" The major answer is that after trying a few different things, the 6-mile maps look aesthetically pleasing at this scale; like, there are decisions for the PCs to make, but not overwhelmingly so, in the travel scale of daily turns. Also the idea of "Dark Ages" means that there isn't much central control in the first place. Adventures or sackings can take place without expecting rescure from a king.

  2. Do you envision your villages, castles, and cities being further apart overall (allowing for more "wildlands" between each settlement.
    Or is it packed in, roughly equivalent to historic levels, but more untamed lands in between these cluster (Points of light)

    1. Good question. More the latter; village fiefs about like medieval England, but fewer of them.

  3. I always like to caution that folks pay particular attention to exactly what is being modeled. Delta's example is specifically based on Medieval England, and I'm sure it is very useful for that type of setting. The exact same exercise using data from Switzerland, Iceland, Mongolia, Cambodia, Peru, etc etc. could and often would give very very different results. If you are setting up a fantasy land that you want to be realistic, you should take into account the types of subsistence and cultural expectations. Iceland for example had almost no villages at all until well into the 19th century. Just something to think about.

    1. Great point! Part of the reason I used England is that there simply seems to be more data available for it (largely due to the Domesday Book, it seems).

  4. Also quite useful for demographics: _Britain Before the Norman Conquest_, from Ordnance Survey (Southampton, UK, 1973), and _Britain in the Dark Ages_ (same publisher) .....allan

    1. Great, thanks for that! Put it on my reading list.

  5. I run a game with higher population densities (being based on the 16th and 17th centuries) and with slightly different sized hexes (the smallest being 2.5 miles on a side and each scale increasing by a factor of 3); but these are numerical details; the methodology is spot on and I may have to steal it when making my maps. Thank you.

  6. Fun project: starting from a random place in a given hex, what is the average distance you will have to travel through the hex to reach a village, a castle, or a town? And if we can find the variance on such things, can we construct a better random encounter table when traveling through an unmapped hex?

    1. Good question. I think you got a pretty good result from asking this on Facebook, yes?

  7. This certainly wouldn't be homogenous across the whole world, would it? On Earth, there are vast regions that have been uninhabited for millennia, or sparsely inhabited through various periods, or inhabited in great numbers for a short period before a civilization moved on or burned out. Does your approach address any of these concepts in a fantasy world or is it generally assumed that there will always be so many villages every few days of travel?

    1. Well, that's a good question, and obviously it's not intended as prescriptive of everything. Maybe along the lines of "where the people are" or "a place like Europe".

      Perhaps the main motivation is that many published fantasy world are far too sparse everywhere. and this is some attempt to boost the civilized lands population up to something that isn't just incoherently empty.