The effective diffusivity of O-2 inside immobilised cell particles has been much discussed. Most reported estimates are based on fitting a mass-transfer reaction model to measured total oxygen uptake rates. The particle diameter has the largest single influence in such models, but its accurate measurement has probably received insufficient attention. We have studied sorbitol and glucose oxidation by cells of Gluconobacter suboxydans entrapped in calcium alginate gel beads. These beads were found to shrink rapidly in air, so that size measurement under water is essential. By comparison with rigid particles of similar known size, it was shown that measurement of the microscopic image gives a systematic underestimate. In consequence, the fitted oxygen diffusivity will be around 20% too low. Careful attention to size measurement gave good agreement between diffusivity estimates from beads with different mean sizes and cell loadings, with a best value of 2.51 x 10(9) m(2)s(-1), 92% of the value for pure water. The estimated diffusivity is not significantly affected by a distribution of bead sizes with up to 10% standard deviation about the same mean.