Interest in the scaling approach to problems of biological design has increased dramatically in the past few years. But thus far no systematic attempt has been made to review the possible pitfalls attendant upon this approach. As a beginning, the problems which can arise from rounding exponents, or taking standard errors at face value, or expressing dependent variables in ratio form are discussed. There follows a discussion of fitting specific functions to scaling data, of the special needs for documentation and of the potential value to be derived from suitable computer programs in scaling studies. Finally, the possible difficulties of demonstrating global optimization in biological systems, the risks of dimensional analysis and the value and nature of scaling models are discussed.