In this study, a versatile approach was presented by using a feedforward multi-layer perceptron (MLP) neural network utilizing Bayesian Regularization and Levenberg-Marquardt training algorithms with the aim of determining the moisture content of wet and dried nixtamal after the application of alkaline cooking. Two different corn varieties were dehydrated at different Ca(OH)2 concentrations, cooking and steeping periods. The corn variety and processing conditions were accepted as inputs of an artificial neural network. In predicting the moisture content of the wet nixtamal, it was discovered that the network where Bayesian Regularization training algorithm was used and which had been designed to contain 20 neurons each in its first and second hidden layers, fitted best with the experimental data. In predicting the moisture content of dried nitamal, the network where the Bayesian Regularization training algorithm was used and which had been designed to contain 40 neurons each in first and second hidden layers, fit best to the experimental data. These configurations can predict the moisture content of wet and dried nixtamal with a regression coefficient of 0.99. Additionally, statistical analysis showed that the most effective two input variables related to the moisture content of corn were corn type and the cooking period.