The research explores whether the scales of natural disasters in Taiwan are distributed in terms of a power law, and if they are, explains why. A power law is defined as an inverse relationship between ranks and scales of objects under consideration. Mathematically, this relationship can be depicted as R(x)=ax-b, where x denotes scales in terms of people killed and wounded resulting from disasters; R(x) denotes ranks with one as the largest scale; a symbolizes a constant; and b is an exponential component. Empirical findings showed that the scales of the natural disasters in Taiwan over the past 100 years indeed follow a power law distribution. Under the observation that the distribution of scales of natural disasters fits a power law, we investigate further its implications and explore possible policies that could change the slope and position of the power law, in order to mitigate disastrous damages. We further suggest that increasing disaster mitigation costs for a certain type of disasters while maintaining the total cost of disaster mitigation will not reduce the disastrous loss and that the government should increase disaster mitigation costs for all types of disasters. The research verifies the universality and invariability of power law in the case of scales of natural disasters. The results could provide insights into disaster mitigation policies.