Abstract: The debris flow deposition is the final product of debris flow. It contains a lot of information of transit process and development of debris flows. The research of deposition has reflected the debris flow hazard and activity intensity. Consequently, solid grains in the debris-flow deposits display a marked self-similarity in geometrical shape and scale-invariance in size according to fractal theory. The particle fractal dimension of debris flow deposition is calculated with fractal theory. By analyzing the relationship between fractal dimension and the length of the main channel, the ratio of loose material length along the channel to the total channel length, average gradient of the main channel, and maximum elevation difference, this paper finds the nonlinear response between fractal dimension and its influencing factors. According to fractal dimension data of debris flow in the Wudongde reservoir area, taking the aforementioned influencing factors as the input units, the support vector machine forecasting model is established. The fractal dimension of debris flow deposits is predicted. The maximum error is only 1.25%,which means the predicted values are consistent with the measured values. It is indicated that the support vector machine model has a great fitting and generalization ability. It is an effective method for prediction of debris flow deposits fractal dimension, which can also be used to deduce the distribution law of particle for debris flows without sample sieve analysis. Besides it provides a new idea for researching the following subjects: formation mechanism, type, hazard of debris flow and formation characteristic, physical and mechanical properties of debris flow deposits.