Backbreak is an undesirable phenomenon in blasting operations, which can bedefined as the undesirable destruction of rock behind the last row of explosive holes. To prevent and reduce its adverse effects, it is necessary to accurately predict backbreak in the blasting process. For this purpose, the data obtained from 66 blasting operations in Gol-e-Gohar iron ore mine No. 1 considering blast pattern design Parameters and geologic were collected. The Pearson correlation results showed that the parameters of the hole height, burden, spacing, specific powder, number of holes, and the uniaxial compressive strength had a significant effect on the backbreak. In this study, a multilayer perceptron artificial neural network with the 6-12-1 architecture and six multiple linear and nonlinear statistical models were used to predict the backbreakin the blasting operations. The results of this study demonstrated that the prediction rate of backbreak using the artificial neural network model with R2 = 0.798 and the rates of MAD, MSE, RMSE and, MAPE were0.79, 0.93, 0.97 and, 11.63, respectively, showed fewer minor error compared to statistical models. Based on the sensitivity analysis results, the most important parameters affecting the backbreak, including the hole height, distance between the holes in the same row, the row spacing of the holes, had the most significant effect on the backbreak, and the uniaxial compressive strength showed the lowest impact on it.