Steffo,

You can use the

FlipCalc I made to determine the odds of a flip, if that will help.

You can see that having captured a city makes it much more likely to flip, most flips will occour in occupied territory, not in uncontested border cities.

This makes sense to me. You capture a city and while there is a large active population they have a good chance of flipping. As your troops squash the resistance that chance drops. As you secure other cities bordering it that chance drops. Finally, over time the citizens are assimilated in to your culture and the chances drop to zero.

In a non-captured city of the target civ your only chance of flipping it is if you are very strong culturally. Even if the city is 4 times closer to you capital then the target civ's, your civ has 3 times the culture of the target civ, you only get a 0.6% chance of flipping it per tile you own in their 21-tile radius. If they have units garrisoned there that drops by 0.2 per unit.

I have (before) had this situation where I had 6 of their tiles. That gives a 3.6% chance of flipping, not including the -0.2% for each unit there. After 30 turns or so it was time for war anyway, so I didn't get the city.