You only need 3 castles.

[ScrapTower]

One for Infantry, one for Cavalry, and one for Missiles. You might want one more in the Holy Land, and maybe one for gunpowder. Its too expencive to tech up more castles then you actually need. I think Im going to start playing this way. What do you think?

[IPoseTheQuestionYouReturnTheAnswer]

Yup, I always maintain an overwhelming city:castle ratio, probably somewhere in the 5:1 ratio. I usually make one in each hotspot for fighting, you really don't need any more than that.

Oh, and the gunpowder infantry units are trained in cities anyway, so you don't need to dedicate a castle to them. Cannon's don't take long to access anyway once gunpowder arrives.

[Daveybaby]

True, except for:

(1) Excessive distance from your castles to the front line makes getting new troops to the front, not to mention retraining old beaten up ones, take a looong time - which can put you at a significant disadvantage.

(2) Theres a limited number of units you can build per turn in a castle, and a limited number of each type of unit become available each turn, per castle.

[Sheogorath]

I prefer to establish my castles in rings, covering choke points (EX: The start of Iberia, Marakesh, etc. etc.), as well as one or two 'inland' castles to supply real military units to my big cities if need be.
You dont get as much money as you normally could, but usually my inner castles are landlocked and dont produce that much income as cities anyway.
I find this also rapidly bogs down any AI attack. They might get past your first ring, but then they have to bring up reinforcements because attacking your first ring of castles wore them down so much.

[Flavius Gonzo]

This is a major dilemna in my English campaign too. In addition to the timing issue though, which you can account for with sufficient planning, there is also the cost issue. If you have a unit that costs 250 to upkeep but the castle is a 7 turn walk/ride from your front lines, that means it really costs an add'l 1500 per unit to get them anywhere they can see action. Yikes.

So, I think there's value in keeping at least some castles that you expand especially inland ones that wouldn't be earning giant revenues as a town.

But, as your borders grow, it's probably a good idea to start flipping most far away castles to towns, although I don't feel comfortable yet with how many/how often. (For example, I struggled last night with the decision to flip Cannes, where I had made big military investments earlier in my campaign.)

[Bob the Insane]

It really is a good dilema... You don't want to lose you old castles as all your best troops are trained there, but you are running out of money to train them or build the better buildings. But as you empire expands those old castles get further and further from the borders making them awakward or training or retaining units.

In dangerous areas a castle is much easier to defend (or at least beat off the assaults) and you can retrain your troops there. But this takes more money and a city can produce cash and allows for a garrison for free. But the free garrsion is unlikely to hold off anything but the lightest assaults unless you have access to advanced militia units, but by that poin the dynamic of the game is changing to favour cities even more. But you would never have gotten to that point without your castles...

Wow...

Personally I have about 3:1 ratio and while money is aways tight, training and retraining is usually available and my lands are well defended...

[Lochar]

Just remember you can only be offered certain guilds to castles only.

Swordsmith, woodsmen, and a few others. You could be preventing self from some knights guilds or other types.

[Aenlic]

With limited numbers of troops available each turn per castle, it's probably a good idea to have at least 4 castles. Infantry from two of them (since you use more infantry than cavalry per army), cavalry from one and a fourth producing missile troops and siege engines. As your empire grows, you can double this number and use the extra set for retraining and resupply of existing troops so you don't have to give up a production slot otherwise used for new troops.