This paper describes the statistical properties of the mixed diffusion-jump process and tests how well it explains the distribution of Korean stock returns. The parameters of the mixed diffusion-jump process were estimated using maximum likelihood estimation for the composite stock price index and 20 individual stocks, and a Kolmogorov-Smirnov test was conducted as a goodness-of-fit test. The study demonstrated that a jump process exists in the generating process of Korean stock returns and established that the mixed diffusion-jump process is a candidate comparable to the mixed normal distribution, which had previously been considered the most promising candidate for explaining the distribution of Korean stock returns. The implications of these findings were also discussed.