A stochastic inverse technique based on Gray-code genetic algorithm (GGA) is proposed for inversion of particle size distribution from dynamic light scattering (DLS) data. It is shown that GGA inverts particle size precisely for different distribution with no random noise. GGA shows stability for distribution with random noise and gives distribution of main peaks. In numerical experiments of latex sphere, GGA inverts dynamic light scattering data with biomodal distribution successfully. Compared with standard genetic algorithm and inverse Monte Carlo method, GGA shows high efficiency in searching optimum results.