The safety assessment of the production process can provide scientific basis for management decisions. The application of neural network theory in system safety assessment can overcome some traditional deficiencies of traditional safety assessment methods and arrive at accurate assessment result. The paper introduced Genetic Algorithm (GA) into the optimization of BP algorithm, which use GA to optimize the weight and threshold of BP and then utilize BP network to seek for exact solution, so that the target of global optimization and fast efficiency can be achieved. Based on this, the GA optimized neural network was applied in the system safety assessment, and the assessment model based on the algorithm was proposed. The example to verify the effectiveness of GA optimized BP algorithm was also provided.