Regularity of RepeatingTimer interrupts?

[Abrakadabra]

How accurate can I count on the trigger timing of ISRs registered with RegisterRepeatingTimer() to be?

I have the idea of using a repeating timer to do regular velocity calculations on the two quad encoders attached to my gear boxes. Basically the ISR would read the encoder counts, subtract the previous counts from each to get the differences, and divide by the repeating timer interval (most likely 256ms, so I can divide using bit shifting). The resulting velocities would be stored in global variables to be accessed by my PID code running in the "default" loop.

Will this work?

[BradAMiller]

The code resets the timer value as the first thing when the interrupt occurs, so it should stay pretty accurate. I would think it would be close enough for the calculations that you need to do.