The dynamics of a bacterial population exposed to the minimum inhibitory concentration (MIC) of an antibiotic is an important issue in pharmacological research. Therefore, a novel antibiotic susceptibility test is urgently needed that can both precisely determine the MIC and accurately select antibiotic-resistant strains from clinical bacterial populations. For this purpose, we developed a method based on Fick's laws of diffusion using agar plates containing a linear gradient of antibiotic. The gradient plate contained two layers. The bottom layer consisted of 15 mL agar containing the appropriate concentration of enrofloxacin and allowed to harden in the form of a wedge with the plate slanted such that the entire bottom was just covered. The upper layer consisted of 15 mL plain nutrient agar added with the plate held in the horizontal position. After allowing vertical diffusion of the drug from the bottom agar layer for 12 h, the enrofloxacin concentration was diluted in proportion to the ratio of the agar layer thicknesses. The uniform linear concentration gradient was verified by measuring the enrofloxacin concentration on the agar surface. When heavy bacterial suspensions were spread on the agar surface and incubated for more than 12 h, only resistant cells were able to form colonies beyond the boundary of confluent growth of susceptible cells. In this way, the true MIC of enrofloxacin was determined. The MICs obtained using this linear gradient plate were consistent with those obtained using conventional antibiotic susceptibility tests. Discrete colonies were then spread onto a gradient plate with higher antibiotic concentrations; the boundary line increased significantly, and gene mutations conferring resistance were identified. This new method enables the rapid identification of resistant strains in the bacterial population. Use of the linear gradient plate can easily identify the precise MIC and reveal the dynamic differentiation of bacteria near the MIC. This method allows the study of genetic and physiological characteristics of individual strains, and may be useful for early warning of antibiotic resistance that may occur after use of certain antimicrobial agents, and guide clinical treatment.