In today's digital era, many industries are affected by digitization. One of them is the newspaper industry which causes demand uncertainty and is difficult to predict. This causes a very high rate of return or newspaper returns which will result in losses for the company. Therefore, it is necessary to use a newspaper demand forecasting method that has a low error rate. The method used is trend line analysis method and backpropagation neural network. The data used is actual demand data, sales results, sales prices and stock from January 2019 to January 2020. The calculation process uses Microsoft Excel and Matlab. In addition, the selection of the best method is obtained by comparing the smallest MSE value. Based on the MSE comparison, it was found that the forecasting method with the smallest error rate was the backpropagation artificial neural network method with an MSE value of 0.0104 Therefore, the best forecasting method for forecasting the demand for the number of newspapers PT. Solopos script is a backpropagation method for artificial neural networks