Another guesstimate method I've used in the past is to have an average # of people per building (O=occupants). Then divide the population (P) by O, and you've got an estimate of how many buildings. From there, knowing how big the average building is, you can get a decent estimate.
I've used an estimate of 5 people per building (anachronistic view of the in-town family, ma, pa, 3 kids living above the shop). Obviously, barracks, inns, apartments, would have different sizes and O values.
I haven't seen the MMV series, but it sounds like it has a pretty thorough model to use.
Janx