The 5e DMG has no size category for cities larger than 25,000 people: because cities require so much surplus food, such cities are “very rare”. Similarly the 3e DMg says that a 100k metropolis should be the exception, not the rule. This is bad advice. Every game world should be dotted with metropolises of mind boggling scale. Screw medieval demographics.
DNS basically has three adventure settings: dungeon, wilderness, and urban. D&D dungeons are not pokey prisons beneath a castle. They are unmapped mega-labyrinths and Mythic Underworlds. Who cares why the dungeon was built or where monsters’ food comes from? D&D wilderness is not a cozy Sherwood Forest where every encounter is with a jolly bandit in Lincoln green. It’s more like that Oregon trail game where you always die of cholera but instead of cholera it’s dragons. It’s the Mythic Wilderness. Similarly D&D urban adventuring shouldn’t be restricted to plausible little 25k cities – the same population as modern Port Chester, NY. The players don’t play D&D to explore fantasy Port Chester. The great D&D cities are Greyhawk, Waterdeep, Lanhkmar – all aspects of the Mythic City – filthy, vast, unmappable.
In practice I’ve found that the metropolis is the standard urban adventuring setting. When you settle down to an extended urban story arc, you do so in a city big enough to stretch your arms in. D&D is not a research paper or a movie. It doesn’t require historical plausibility and it doesn’t cost you anything to build sets. D&d should embrace that. Instead of advising prudent and conservative little settlements, it should recommend vast old sprawling cities containing stinking treasure-laden rivers, ancient forgotten palaces jutting between slum roofs, armies of beaurocrats in competing courts, entire neighborhoods which only appear under certain moons, and strange monstrous denizens who remember ancient days.