We have developed a hybrid optimization algorithm for binary inversion of geophysical data associated with lithologic inversion and time-lapse monitoring. The binary condition is designed to invert geophysical data for well-defined physical properties within discrete lithologic units, such as a salt body within a sedimentary host, or temporal changes in physical property associated with dynamic processes, such as the density and conductivity change in an oil and gas reservoir or groundwater aquifer. The solution of such inverse problems with discrete model values requires specialized optimization algorithms. To meet this need, we develop a hybrid optimization algorithm by combining a genetic algorithm with quenched simulated annealing. The former allows for easy incorporation of prior geologic information and rapid buildup of large-scale model features, whereas the latter guides genetic algorithm to faster evolution by rapidly adjusting the finer model features. In examining the performance of the hybrid algorithm, we note its superior performance. Our investigations of the algorithm's capability with a 3D gravity inversion for the SEG/EAGE salt model and a time-lapse gravity data set from an aquifer storage and recovery process reveal its benefits.