Edwardsiella tarda, a bacterial pathogen, causes opportunistic infection to fish species and it brings about a hemorrhagic septicemia, called edwardsiellosis. It leads to high mortality and severe economic losses in farmed fish. The control of infectious disease in the fishery is a critical issue linked with the economy. While several technologies are used to detect and control infectious diseases in fish farm, these diseases are still a cause of economic loss. Therefore, the need for a system to quickly detect such diseases is increasing. Accordingly, we sought to quantify and observe the behavior of fish in respect to bacterial infection. Self-Organizing Map (SOM) was utilized to reveal fish behavior in response to bacterial infection in zebrafish, Danio rerio. In order to observe the behavior of zebrafish, we assembled an equipment system for a video recording. Bacterial inoculation strategy was achieved by immersion to bacterial suspension. Zebrafish behavior was recorded for an hour daily. Through the recording result, we determined behavior characteristics based on parameters, such as speed, acceleration, stop duration, stop number, locomotory, turning and wandering. In consequence of analysis, the gradient was mainly observed according to a dose dependent manner and it showed the different behavior patterns between experimental group and control group. This difference of behavior change was significantly distinguished by 6 clusters. This study of fish pathogenesis with SOM is suitable in addressing host property and in establishing reference systems for bacterial infection. We aimed to identify and characterize fish behavior for infection by E. tarda and it will also provide a basic framework for analyzing the behavior of fishery with regard to bacterial infection. Furthermore, it can be applied to establish a form of a bio-indicator for disease control, as well as prevent economic loss due to infectious diseases.