Ethylene and propylene separation from the mixture produced by steam cracking of saturated hydrocarbons is a highly energy-intensive process. In this, plant operators face several problems due to changes in feedstock, loss of ethylene and propylene with other streams in intermediate columns, varying demand for ethylene and propylene, and higher utility consumption. Plants need to adjust the operating conditions to maximize profit while meeting the product demand and specifications. In order to maximize profit, it is important to reduce losses as well as energy requirements, which are conflicting and require multiobjective optimization (MOO). In this study, cold-end separation process of an ethylene plant is simulated in Aspen Hysys, and the simulation model is validated with typical design data. Then, using this model, MOO of the cold-end separation system is studied using the elitist nondominated sorting genetic algorithm. It is shown that the plant can be operated at different optimal conditions, each of which involves some trade-off among the objectives of interest.