Random and directed motility of bacterial populations were assayed by monitoring the flux of bacteria through a microchannel plate (a porous glass plate comprising a fused array of capillary tubes) separating two identical stirred chambers. Cells, washed free of growth medium by a new filtration method, were added to one chamber at a low density. Their number in the other chamber was determined from the amount of light scattered from a beam of a laser diode and recorded on a strip chart. Diffusion coefficients were computed from fluxes observed in the absence of chemical gradients, and chemotaxis drift velocities were computed from fluxes observed in their presence. Cells migrated through tubes of diam 10 microns more rapidly than through tubes of diam 50 microns, suggesting that the straight segments of their tracks were aligned with the axes of the smaller tubes. Mutants that are motile but nonchemotactic could be selected because they move through the microchannel plate in the face of an adverse gradient. Weak chemotactic responses were assessed from ratios of fluxes observed in paired experiments in which the sign of the gradient of attractant was reversed. Studies were made of wild-type Escherichia coli and mutants that are nonmotile, tumblely, smooth-swimming, aspartate-blind, or defective in methylation and demethylation. Chemotaxis drift velocities for the latter mutants (cheRcheB) were quite small.