Novel methods in molecular biology and advanced technologies have given pharmaceutical research laboratories the capability to test combinatorial libraries rapidly against large numbers of potential targets. Methods to identify optimal assay conditions efficiently are very useful in the development of robotic screening assays where there are numerous variables and potential interactions between the variables. This review discusses the use of statistical experimental design in the development and optimization of high-throughput screening assays. The authors provide a brief introduction to the theoretical basis for experimental design and discuss practical aspects of using these methods in a research laboratory. Two case studies demonstrate the power of the method in solving problems in assay development and illustrate the diversity of potential applications.