Livestock has an important role in the structure of the rural economy, where small livestock is an ideal choice to support the local economy and ensure food security. In Lalat Regency, there is no accurate prediction system to find out the trend of small livestock populations in the future, both increasing and decreasing. To answer this need, this study uses the Artificial Neural Network (JST) technique with the Backpropagation method to predict the small livestock population in the region. This study develops a JST-based prediction system using available training data, training targets, and small livestock population test data. The prediction results for 23 sub-districts in 2023 show that in the first experiment, 18 sub-districts are in accordance with the prediction and 5 sub-districts are not in accordance with the prediction; In the second experiment, 17 sub-districts were compliant and 6 sub-districts were not compliant; And in the third experiment, 21 sub-districts were compliant and 2 sub-districts were not compliant. The prediction process was carried out with a target error of 0.00001, with each experiment using 20 iterations in the first experiment, as well as 4 iterations in the second and third experiments, and the training time was recorded for 00:00 seconds. This system was successfully developed using Matlab software, provides accurate prediction results of small livestock populations, and can be used by the Langkat Regency Agriculture and Food Security Office for better planning in the future.