Implementing the derivative in C/C++

How is the derivative of a f(x) typically calculated programmatically to ensure maximum accuracy?

I am implementing the Newton-Raphson method, and it requires taking of the derivative of a function.