Escalating stringent environmental regulations on fuel coupled with decreased quality of incoming crude oil make energy conservation techniques in fuel production a necessary venture. In this work, a nonlinear programming (NLP) model was proposed to optimize the hydrotreating (HDT) system coupled with a fluid catalytic cracking (FCC) unit in refineries. Two optimization models, i.e. a total sulfur removal (TSR) model and a sulfur compounds removal (SCR) model, were established based on the proposed NLP model that embeds the desulfurization kinetics for the total sulfur removal and the sulfur compounds removal. The effects of operational conditions on both utility costs and impurity removal degrees were investigated. An actual refinery was taken as a case study. Results show that when the sulfur content in the refined vacuum gas oil is 1510 mu g.g(-1), the utility cost of the HDT system can be reduced by 3.75% using the TSR model, whereas it can be reduced by 5.83% using the SCR model. Further reduction of the utility cost in the HDT system can be achieved by taking the sulfur compound removal into consideration. Moreover, the effects of allocation of sulfur in the FCC unit were also investigated by using the SCR model. Results show that the annual cost of the hydrogen consumption can be reduced by 17.31 x 10(6) CNY (China Yuan) and the utility cost can be lessened by 7.29%. For precise sulfur management and effective reduction of energy and hydrogen consumption of the HDT system in a refinery, the operating features of the HDT units coupled with the FCC unit and the characteristics of sulfur compound removal in the hydrotreaters should be considered simultaneously.