Predictive mathematical models of the interactions of a genetic network can provide insight into the mechanisms of gene regulation, the role of various genes within a network and how multiple genes interact leading to complex traits. However, identification of the parameters and interactions is currently a limiting step in the development of such models. This work reviews the state of the art for design of experiments in biological systems and demonstrates the need for improved design of experiments through the use of a model system. Appropriate design of experiments has a profound impact on the ability to identify a model and on the quality of resulting identified model. Key issues include the selection of appropriate input sequences (e.g. random, independent multivariate inputs) and the selection of the sampling frequencies. This work demonstrates that these issues are especially important in the identification of biochemical networks and that the traditional biochemical approach is incapable of truly identifying the behavior present in such networks.