The objective of this research was to estimate the production of sugarcane by using artificial neural network algorithm. The training was done at soma variety of iteration, those are 1000, 2000, 3000, and 4000 with the nodes at the hidden layer. 2,4, 6, and 8. Based on the simulation. it showed that the more nodes and iteration could reduce sum of squere error. for example iteration 1000 at node 2, sum of square error was 0.0148. and at the node 8, sum of squares error was 0.000196 In ither side at the node 2, the iteration 1000, the sum of squares was 0.0148, and the iteration 4000, the sum of squares was 0.00345. The result of the model test showed that sugarcane productivity had the exact for prediction that was from 87. 5 % to 97.15 %, for sugarcane production it had low prediction that was 40.0 % to 92 5 .% and respectively, it had 94.2 % to 98.8 %.