Pro-convulsant agents can cause a series of pathologic swimming behaviors in zebrafish, of which rapid “whirlpool-like” swimming is a representative seizure characteristic behavior. However, there is a lack of good and accurate method to quantify this type of swimming feature. Therefore, in this study, we attempted to establish a new larval zebrafish seizure analysis technique for quantitatively exporting the number of swirly swimming in a certain period. The MARGO software package based on MATLAB was employed to track the zebrafish swimming route and to produce the space–time coordinates. An innovative algorithm was developed based on the factors of vector angle, trajectory length, and swimming duration. Through the optimization of the algorithm, ZebVortrack, an analysis system, was established for automatically recognizing seizure-like swimming behaviors in larval zebrafish, treated with pentylenetetrazol (PTZ). Different pro-convulsant agents, and antiseizure drugs were used to evaluate the accuracy of the ZebVortrack system for larval zebrafish. In a word, we developed a new behavior-analysis system, ZebVortrack, for quantitatively identifying the swirly swimming behavior of larval zebrafish that can automatically and accurately determine the vortex numbers and seizure latency of larval zebrafish in microplate wells so as to provide a new analysis method for study of seizures and drug screening.