To improve the prediction accuracy of concrete dam deformation, a Long Short-Term Memory(LSTM) network-based concrete dam deformation prediction model is proposed, which hasthe merits of excellent nonlinear data mining ability and the long and short-term prediction performance of time series.Example analysis shows that, compared with the commonlyused stepwise regression and multiple regression methods, the LSTM network-based deformation prediction model can effectively mine the complex nonlinear relationship between dam deformation and influencing factors. The modeling and predicting accuracy of the model can be significantly improved, providinga new method for dam deformation prediction.