In this paper, we propose a new three-arm parallel design to investigate biosimilarity between a biosimilar product and an innovator biological product by using relative distance based on the absolute mean differences. In the proposed design, one arm is for the biosimilar product and the other two arms are for the innovator biological product. The distance between the biosimilar product and the innovator biological product is defined by the absolute mean different between two products. Similarly, the distance between the innovator biological products from two difference batches is defined. The relative distance is defined as the ratio of the two distances whose denominator is the distance between the innovator biological products from two different batches. In the proposed design, if the relative distance is less than a prespecified margin, we claim that the two products are claimed to be biosimilar. The statistical test based on the ratio estimator and the linearization method are developed to assess biosimilarity. The power functions of two tests are derived in large sample and compared numerically. Because the statistical test based on the ratio estimator is more powerful than the linearization method, we recommend the statistical test based on the ratio estimator. Copyright © 2012 John Wiley & Sons, Ltd.