Abstract: Isothermal hot compression experiments of HNi55-7-4-2 alloy were conducted with a Gleeble-3500 thermal simulator at deformation temperatures of 873, 923, 973, 1 023, 1 073 K and strain rates of 0.01, 0.1, 1, 10 s^-1, and the high-temperature flow behavior of the alloy was studied. The constitutive model of the alloy was established by the modified Johnson-Cook (M-JC) model and back-propagational artificial neural network (BP-ANN) algorithm with experimental data. The prediction accuracy of two models was comparatively analyzed. The results show that the flow stress of HNi55-7-4-2 alloy increased with increasing strain rate or decreasing deformation temperature. The average absolute relative errors between true stress predicted by the constitutive model on the basis of M-JC and BP-ANN algorithm model and test results were 14.63% and 0.35%, respectively and the correlation coefficient were 0.978 7, 0.999 9, respectively. The constitutive model by BP-ANN algorithm had higher prediction accuracy, and could decribe the high-temperature flow behavior of HNi55-7-4-2 alloy well.