The artificial neural network and multiple regression models have been developed to predict the evenness of cotton ring yarn with process parameters such as front roller speed, spindle speed, nip gauge, back draft zone time and roving twist. The efficiencies of prediction of the two models have been experimentally verified, and the predicted evennesses of cotton ring yarns from both the models have been compared statistically. An attempt has been made to study the effect of process parameters on yarn evenness. The MSE and mean absolute error of ANN modelare lower than that of multiple regression model. The results show that the performances of prediction of ANN models are more accurate than those of multiple regression models.