In order to contribute to the research and development of adhesives for the footwear industry, this paper aims to develop a model capable to predict and optimize the peel strength from the composition of adhesives. The proposed approach is based on three stages: experimental planning of measurements, global sensitivity analysis for uncertainty propagation, and optimization procedure. The design variables are the weight percentages of the solid raw material constituents such as polyurethane, resins, and additives of the adhesive joint. Considering the experimental results obtained for Taguchi design points as input/output patterns, an artificial neural network (ANN) is developed based on supervised evolutionary learning. Using the developed ANN a global sensitivity analysis procedure is implemented and the variability of the structural response of adhesive joint is studied. The optimal solution for adhesives composition for maximum peel strength is investigated based on ANN model and using a genetic algorithm. The proposed approach is able to predict the optimal peel strength including its sensitivity to uncertainties. The results show that the sensitivities of design variables belonging to polyurethane and additive groups are important for optimal adhesive joint. The optimal peel strength based on proposed approach is consistent with the experimental testing data.