Several methods for dispersion of soil were tested for possible use in procedures for extraction of bacteria. Physical cell damage on cells and efficiency in extraction of indigenous cells from soil, were investigated. Cell damage by the dispersion methods was investigated by measuring the physical cell integrity and viability of pure cultures of Escherichia coli and Bacillus subtilis, as well as soil bacteria extracted from soil, when dispersed in slurries of γ-sterilized soil. Separation of bacteria and soil particles on the basis of buoyant density was conducted with the nonionic density gradient medium Nycodenz. When slurries of γ-sterilized soil with added pure cultured cells were centrifuged (10000 × g) over cushions of Nycodenz (1.3 g ml −1), practically all the added cells were recovered in a layer on top of the cushion. This proves that a reversible attachment and cosedimentation is not an important phenomenon in this procedure. The efficiency of the different dispersion methods for the extraction of indigenous soil bacteria, was assessed after separation of dislodged and attached soil bacteria. This separation was done either on the basis of sedimentation rate by low speed centrifugation, or buoyant density by Nycodenz density gradient centrifugation. The physical dispersion by ultrasonic treatment and chemical dispersion by the use of a chelating agent together with a detergent, were inferior to physical dispersion either by Waring blender (for large volumes) or a rotating rubber pestle treatment (for smaller volumes). The physical dispersion did not appear to be destructive to the cells tested.