Ribonuclease, insulin, cytochrome C, myoglobin and ovalbumin were introduced into solutions from which ferritin and lysozyme crystals were grown. These measurements were also performed for the ferritin dimers trapped by growing ferritin crystals. The crystals were later dissolved in a pure solvent, the impurity concentrations were measured by high performance liquid chromatography and the effective impurity distribution coefficient, K, was evaluated relative to the initial concentrations of ferritin or lysozyme. The density of impurity species in crystal relative to its density in mother solution were used to calculate volumetric distribution coefficient, k. These distribution coefficients were found to exceed unity (k > 1) in terrestrial condition for all impurity species, except for insulin and cytochrome C lysozyme. For ferritin dimers, K = 4, k = 1.8 x 10(3). Crystals grown in space under the otherwise identical conditions incorporated lower amounts of all of these impurities, majority of them below the detection limit. The lower impurity incorporation obtained in stagnant solution may be partially due to more difficult impurity supply through the impurity depletion zone arising around the growing crystals at k > 1 in the absence of buoyancy driven convection or stirring. Analytical estimates of the depletion zone show reasonable agreement with measurements for ferritin dimers. Step bunching and other flow-dependent surface processes may also contribute to lower distribution coefficient.