The similarity principle can be stated as: similar things will behave similarly. For example, we believe that two patients with similar baseline characteristics will have similar responses to a drug. The similarity principle is a hidden fundamental principle for all science including statistical science (the frequentist and Bayesian paradigms). Knowing this principle is important in study rare disease from a statistical perspective. Many controversies in rara disease trials, personalized medicine, multi-regional trial, and general statistics, such as Simpson's paradox and multiplicity issues, are rooted in the similarity principle. We will discuss the controversies in the rare disease and personalized medicine, how to use the similarity principle to resolve the issues and compare this innovative approach to traditional frequentist and Bayesian approach.