An agar diffusion method for the assay of colicins A, B, D, E(2), E(3), and K is described. The assays were performed in large, square pyrex dishes that contained an agar layer seeded with an indicator organism sensitive to the colicin. The samples were applied to the agar in steatite beads positioned in a randomized sequence. The plates were stored at 4 C for 24 hr to allow the colicins to diffuse into the agar. After incubation at 37 C, the activity of each colicin preparation was estimated by measuring the diameter of the zone of inhibition of the growth of the indicator strain around each bead. The results of each assay were subjected to a statistical analysis, which included an analysis of variance and calculation of the theoretical regression and the confidence interval of the assay. The size of the inhibition zones, the form of the regression, and the slope of the regression of the responses were affected by the type and concentration of the agar, the depth of the agar layer, the indicator organism, the indicator inoculum density, and the time allowed for prediffusion of the colicins. Optimal conditions for the assay of each colicin were determined. Using a four-point assay design, the relative colicin concentration of unknown preparations was estimated in terms of a standard preparation of the same colicin. The experimental error of these assays (95% confidence interval) was about +/- 10%.