How Do You Assess Whether a Trading Strategy Is Overfitted to Historical Data?

A strategy may perform exceptionally well on the period used to develop it but struggle on new data. What testing methods would you use to determine whether its apparent edge is likely to survive outside the original sample?