In order to solve the problem of large errors in effect quantity prediction caused by small samples in dam safety monitoring, a new approach for prediction of effect quantity, through increasing sample information, was established based on discrete and regression analysis of two-dimensional normal information diffusion. The effect quantity of dam safety monitoring was analyzed and predicted using the basic theory of fuzzy reasoning and information concentration, as well as the grey correlation degree method. The comprehensive effectiveness index was used to evaluate the effect of the information diffusion approach. The results show that the information diffusion approach avoids the difficulties in effect quantity prediction using small samples under the influence of some uncertain factors, and improves the fitting and forecast accuracies.