Concrete-filled steel tube (CFST) columns have features such as reduced construction costs and more dilation and carrying capacity to reinforced concrete columns. Therefore, Concrete-filled steel tube columns have been increasingly adopted in many modern structures such as bridges and high-rise buildings due to the beneficial composite action between steel tube and concrete core. Mainly, experimental and codes models of a limited number of experimental samples have been used to predict the axial ultimate strength of these sections, which are derived from regression analysis. Therefore, the use of prediction models is not widespread in the most reasonable range. In this paper, a novel predicted model of the axial ultimate strength has been developed based on Artificial Neural Network (ANN) using a large experimental data set more 1168 sample data on CFST. The predicted results of the developed ANN model were compared with tow empirical newest models and AISC and EC4 code relations using several error statistics. The proposed Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm improves the root-mean square error and the mean absolute error is about 90% and 30%, respectively, and also with a coefficient of correlation of 0.993, the best correlation between predicted and experimental data is shown in comparison with other prediction models.The results indicate that the successful application of the Neural Network in the estimation of the ultimate strength of these sections has been shown to be experimental and codes models.