An experimental design scheme proposed for process and product development integrates the artificial neural network, random-search algorithm, fuzzy classification, and information theory. An initial batch of experimental data is first collected to construct a neural-network model. Random search generates a number of candidates for the next batch of experiments. A fizzy classification algorithm is used to find the cluster centers of these candidates. An information free energy index is defined to balance the need for better classification and the relevance of each class in optimization. New experiments are performed at these cluster centers to validate the model. The procedure is repeated until an optimal solution is reached. Case studies using a mathematical model and a real industrial pigment-blending project illustrate the abilities of this method to locate multiple optima and handle multivariable experimental design.