Petroleum refineries are major greenhouse gas emitters due to the considerable amount of energy used by some of their operations such as the crude distillation unit (CDU) and the fluidized bed catalytic cracking (FCC) unit. Therefore, energy savings and emission reduction have become very prominent issues in today's petroleum-refining industry. Energy integration offers considerable benefits in terms of reducing energy costs of industrial processes by increasing heat recovery and reducing utilities consumption. This can be achieved by combining all sources for heat and power demand/supply within an individual process or from different processes. Several potential opportunities for improving the energy efficiency, and consequently, reducing CO2 emissions of refining processes have been investigated over the years via implementing heat integration within a single process unit or among different refining processes. However, the tradeoff between CO2 emissions and other economic or operating objectives has not been widely covered. In this study, multi-objective optimization has been implemented for an integrated model of a CDU/FCC complex using a binary-coded elitist nondominated sorting genetic algorithm (NSGA-II) to investigate the tradeoff between CO2 emissions and the economic objectives. Pinch analysis is used for process heat integration of the CDU and FCC units. Two modes of heat integration are used for comparison. Direct integration is considered first, where all the hot and cold streams from both units are used to compose the grand composite curve, and are therefore assumed to be available for integration. Secondly, indirect integration, also referred to as the total site heat recovery, is considered, where different processes are linked to the same utility system. The results on the Pareto-optimal curves and optimal operating conditions are presented and their significant features are discussed.