A system for predicting the availability of food commodities can help in making decisions. Artificial Neural Network is a method that is able to carry out mathematical processes for predicting the availability of food commodities. With the Backpropagation algorithm, the previous data processing is used as input to predict the availability of food commodities. Data processed as input variables are Area of ​​Harvest, Productivity Level, Number of Production and Number of Consumption Needs. While the processed food commodities are types of Rice, Corn, Soybeans, Peanuts, Green Beans, Cassava and Sweet Potatoes. The data was taken from 2006 to 2013. The years 2006 to 2012 were used as input data, while for 2013 they were targeted data. Some stages of Backpropagation are initializing weights, activating, calculating input weights and output biases and changing weights and biases. This stage will obtain the output to be achieved with the smallest error approach so that the predicted results of the availability of food commodities are obtained. The training process uses Matlab software tools 6.1. The result is a prediction of the amount of food commodity availability by the training and testing process producing actual output as the target achieved.