Sieving rate was estimated based on the particle percolation model, which comprises three probability models for (i) a vibrating particle bed, (ii) the boundary between particle bed and sieve Surface, and (iii) sieve openings. It was assumed that the vibrating particle bed is formed front a simple cubic lattice, and the percolation probability of undersized particles throughout the particle bed is represented as a function of porosity. The porosity variation mechanism was elucidated by simulation, and the effects of undersized-particle concentration were elucidated by experiments. Subsequent particle behaviors on the sieve surface were analyzed in order to obtain the passage probability of undersized particles through the boundary between the particle bed and the sieve Surface. Additionally, in order to obtain more accurate passage probability of the undersized particles through the sieve openings, effects deriving from vibration conditions were taken into account. As a result, the estimation equation comprising these three probability models was found to allow prediction of sieving rate under the operating condition that the mass ratio of such undersized particles is 10 percent or smaller.