The purpose of this study is to predict the total end breakage per machine in 40 hrs of cotton yarn in a rotor spinning machine based on yarn count (yarn count), rotor speed (rotor speed), opening roller speed (opening roller speed) and residual trash content in draw frame sliver. . This study uses an artificial neural network (ANN) method in predicting a desired output. Furthermore, the artificial neural network is modeled with several model variations. From several modeling and testing carried out, starting from varying the number of nodes, the amount of alpha, the number of hidden layers, the number of iterations, it can be obtained that the results of using an artificial neural network with 1 hidden layer, 3 nodes, alpha of 0.3 with 50,000 iterations have more optimal results compared to the others because the resulting output is close to the target with an R-squared value of 0.984968. This shows that there is a large or close correlation between the actual variables and the variables in the artificial neural network. The novelty of this study is the use of ANN for the first time in predicting the total end breakage per machine in 40 hrs of cotton yarn in a rotor spinning machine. This method can facilitate top management and especially the Quality Control section in making decisions to set the parameters of the rotor machine in order to minimize the occurrence of yarn end breakage per machine in 40 hrs.