Several methods have been introduced to estimate the optimum level of dietary nutrients such as metabolisable energy (ME), crude protein (CP), and lysine (Lys) in broiler chicken production. Performance optimisation is usually measured as maximising body weight gain and minimising adjusted feed conversion ratio (Adj FCR). One useful method is to model a system that requires an explicit mathematical input-output relationship. Group method of data handling-type neural network (GMDH-type NN) and genetic algorithm (GA) is used to model and optimise an output in an imprecise environment (Yao, 1999). The purpose of this study was to apply the GMDH-type NN and GA methods to provide an optimised formula for broiler chicken performance based on the dietary level of ME, CP, and Lys.