A new target detection method in hyperspectral images based on projection pursuit (PP) and genetic algorithm was introduced. First,skewness and kurtosis which were susceptible to outliers were chosen as the projection index (PI). Then,Real coding based Accelerating Genetic Algorithm (RAGA) was used to find optimal solutions for a selected PI. The proposed method was studied using real hyperspectral data. Experimental results show that the method can effectively and reliably detect the target from hyperspectral images.