The invention discloses an automobile high-frequency noise prediction method based on a correlation interval uncertainty theory. The method comprises the following steps: selecting geometric and material parameters with poor consistency in engineering as interval uncertain parameters on the basis of establishing a whole vehicle statistical energy model and carrying out acoustic package definitionand model benchmarking, and describing a relation between the uncertain parameters through a linear constraint inequation; calculating the local sensitivity of the noise in the vehicle to each uncertain parameter, reordering the uncertain parameters according to the magnitude of the sensitivity, and substituting the sensitivity into an inequality equation to obtain an ordered uncertain parameter vector and an uncertain parameter correction perturbation interval; and carrying out first-order Taylor series expansion on the noise in the vehicle and bringing the noise into an uncertain parameter center value and a correction perturbation interval to obtain a distribution boundary of the high-frequency noise in the vehicle. The correlation interval uncertainty theory is put forward for the first time, the relation between uncertain parameters is described through the linear inequation, and the method has important value in science and engineering.