In order to make a system safer, it is necessary to determine which hazard should be removed by considering the interactions of all hazards in the process. In this study, the influence of the hazards related to iron works in a construction firm is examined by Bayesian network approach. If only the most important hazard is identified and removed from the work environment, it may be erroneous to comment on the levels of the remaining hazards. A Bayesian network is proposed to remove this problem. If there is more time and budget than planned for a process, these capacities will have to be reduced in other processes. Linear programming is proposed to remove this problem. A strong decision support system has been established by combining a Bayesian network and linear programming. A construction firm application has been demonstrated using the proposed method.