The design of an alluvial channel for critical conditions requires information about four basic parameters, i.e., particle size d, water depth y, energy slope Sf and average velocity u. Most of the investigators use a standard or modified form of the critical Shields' parameter to define the incipient motion of a particular grain size. Shields' diagram describes an iterative process of determining the incipient motion parameters. To design a process that does not require iteration, one must know the resistance relationship. The absence of a universal resistance equation increases the difficulty of defining the model. Metamodelling techniques are widely used in engineering designs to simulate a complex system. This article describes a Metamodel, which employs the radial basis function (RBF)network to predict the average velocity u and energy slope Sf based on experimental data of incipient condition. With the help of the Metamodel generated by the RBF network, design curves have been presented to predict the u and Sf based on known values of the particle size and water depth. It has been found that this model improves the predictability of incipient motion.