Sexual dimorphism in body size has received increasing scientific interest during the past decade. In most of the papers on the subject, the magnitude of sexual size dimorphism is expressed as some conversion of body size ratio. There are several possible flaws in using such ratios, however, the most serious one concers the scaling of body size and size differences between sexes. Using ratios may easily lead to wrong conclusions. In this paper we emphasize some fallacies related to ratios and present a less biased method, based on residuals of linear regression. This method is suggested to be used in comparative analyses of sexual size dimorphism.