A primary goal in neuroscience is to understand how the brain gives rise to innate and learned behavior. One way to expand the contributions of zebrafish in this respect is to use a computerized system that can train larval zebrafish in various behavior assays. This will allow more efficient genetic and chemical screens aimed at investigating mechanisms of brain function. In this chapter, we describe a simple, economical and efficient way to automate avoidance learning in larval zebrafish. A webcam is used to acquire images for custom-built software to identify the position of fish in real-time. Based on this information, the computer delivers training stimuli from a universal serial bus module board that connects to the testing tanks through a small electrical circuit board. The training protocol is pre-programmed and the entire process is automatic, so no experimenter needs to be present and no offline video analysis is required.