The axial distribution of gel beads suspended completely in a bubble column was measured by sampling the three-phase dispersion through a vertical pipe located at any height of the column axis under various operating conditions of superficial gas velocity U-G, average gel beads holdup epsilon (S) and unaerated slurry height H-D, The gel beads used were prepared by entrapping glucose oxidase as well as fine palladium particles within calcium alginate gel to catalyze the air oxidation of glucose for efficient production of calcium gluconate. The conventional sedimentation-diffusion model, which includes the gel beads dispersion coefficient E-P and their settling velocity v(P) to describe the axial distribution, is found to give unreasonable E-P values being at least twice as high as than those predicted from the published correlations, and also twice as high as the estimated values of the liquid dispersion coefficient E-L in the three-phase bubble column. This was assumed to be ascribed to the reduced apparent values of v(P), since the intrinsic E-P value should be nearly equal to the E-L value. In view of the fact that the gel bead has almost the same density as the liquid phase and the size is comparable to the mean bubble size, the model was modified by introducing an additional parameter, the rise velocity of the gel beads swarm um representing an effect of rising air bubbles on settling gel beads. The values of um were determined from the observed axial distributions by assuming that the E-P values are given by the reported correlation of E-L in suspension of the inert calcium alginate gel beads, and empirically correlated with U-G and epsilon (S). The axial distributions calculated from the modified model well agreed with the observed ones for the different H-D.