It depends if you want attitude or flight path stabilization. The gyro alone will provide enough information for attitude stab. Consider pitch. An unintended pitch up is sensed by the gyro. Your control system corrects it. The airplane is now on a different flight path. So, pitch is not a 2nd order system, but we already knew that. Acceleration sensors can add the extra info you need. Developing the maths models is the tricky bit!!!

(for the pedants - these comments apply to low cost, low weight, low complexity systems for mini aircraft use - this is not a space shuttle).

Try not to do everything with one micro. I do the sensing and basic processing with a pic. Its fast, cheap, does the duty cycle measurement very easily, and some simple maths and comms. This way the sensor (bit in the servo case) can be made to appear like a sensor on a virtual airplane of predefined dynamic specification. Have one of these for each axis and do the heavy maths somewhere else, using a processor more suited to that task (an arm or similar - plenty of prefab boards to choose from). The big processor deals with the virtual vehicle, so if you change the vehicle dynamics there is no need to change the software here.

C - All my pic code is in C - that way I can figure out what it does!!!

Hobby shops are wrong - I use a gyro on rudder in my jet. There are many apps for gyros other than helis. The model gyros are quite unsuitable for what you want to do. About the only bit you would keep is the sensor!!!

Hope this helps.