Abstract: Dam deformation is usually caused by coupling effects of various factors and it is an uncertain system that contains highly gray characteristic and fuzziness. Because no formula is yet available for deterministic mathematical or physical description of dam deformation influenced by several factors, combination models are more rational than single models in prediction of dam deformation, given that they are more effective in using information from the submodels and better solve those problems complicated and uncertain. Previous studies on such combination models are quite lacking and have no consideration of time factor. This paper presents an artificial neural network combination model that considers time factor and combines ANFIS-GM and GA-BP models for prediction of dam deformation. Application to a core rock-fill dam in Southwest China shows that this combination model not only has a higher accuracy than the submodels but it is more accurate than the minimum prediction error combination model with no time factor considered. Relative to the latter model, it shows an accuracy 10.50 mm higher in average.