For approval of generic drugs, the U.S. Food and Drug Administration (FDA) requires the evidence of bioequivalence in average bioavailability be provided. This is based on the Fundmental Bioequivalence Assumption from FDA that if two drug products are shown to be bioequivalent, it is assumed that they are therapeutically equivalent and can be used interchangeably. Recently, there are a few statistical considerations for assessing biosimilarity. In this article, we propose a new method based on a frequency estimator to evaluate biosimilarity; the large sample properties of the estimator, the power analysis, and calculation of the sample size are considered. Under a three-arm parallel design, the power comparison between the frequency estimator method and the other existing methods is studied through simulation, and the statistical test based on the proposed method is more powerful than for the other two methods. Simultaneously, we compared the relative performance of the three methods, and the empirical consistency and inconsistency probability are obtained in evaluating the biosimilarity.