All risk is of course calculated.
Even with gravitation, for pratical issues Newton's law is fairly used because it is a very good approximation, and a 5cm error when computing the position of the moon is not a big deal.
The thing is, when dealing with a problem one has to evaluate in what regime the problem undergoes. Sometimes the system always stays in a classical regime and it is a very good approximation to use non-relativistic results.
Of course when in doubt use the complete theory and then proceed to evaluate the contribution of relativistic corrections to see of their importance.
For theoretical purposes, always use the complete theory, and then compute in different regimes if necessary. I would say this is a kind of 'common sense' aproach.