The present invention relates to a system and a method for determining a stroke based on a voice analysis. According to the present invention, voice data of subjects are collected to extract and analyze voice onset times to determine stroke patients based on voices. The system for determining a stroke generates and collects voice data from test subjects reading a predetermined word that includes a plosive sound. The system for determining a stroke extracts and calculates voice onset times from the voice data to calculate probability parameters for the voice onset times of each of a normal group and a stroke patient group. The system for determining a stroke uses a set of probability parameters to determine an integration section, and calculates probabilities of being in the normal group and the stroke patient group. The system for determining a stroke applies the calculated probabilities to the Bayes theorem to determine whether the subjects are stroke patients.