Toxic gas leakage in metallurgic plants has emerged with the growth of crude steel production in recent years, causing damage to people, facilities, and the environment. Poisonous gas leakage can lead to other severe accidents including fires, explosions and gas poisoning. In this paper, we propose a risk assessment system (RAS) for toxic gas leakage using a fuzzy evaluation method integrating the entropy weighting method (EWM) and the order relationship method (ORM) and compiled an index system consisting of four first-level indices and fifteen secondary indices. The first-level indices are blast furnace safety performance, protective facilities, evacuation and dilution facilities, and poisonous gas management. The four first-level indices’ toxic gas leak evaluation result is 0.8581, 0.8971, 0.7733, and 0.8652, respectively. We observe that the overall status of the metallurgical plant is “excellent”, yet the result for the evacuation and dilution facilities was less than 0.8, indicating that there is still room for improvement. The risk evaluation time is reduced by forty percent by adopting RAS.