Cold-end systems are heat sinks of thermal power cycles, which have an essential effect on the overall performance of thermal power plants. To enhance the efficiency of thermal power plants, multi-pressure condensers have been applied in some large-capacity thermal power plants. However, little attention has been paid to the optimization of the cold-end system with multi-pressure condensers which have multiple parameters to be identified. Therefore, the design optimization methods of cold-end systems with single- and multi-pressure condensers are developed based on the entropy generation rate, and the genetic algorithm (GA) is used to optimize multiple parameters. Multiple parameters, including heat transfer area of multi-pressure condensers, steam distribution in condensers, and cooling water mass flow rate, are optimized while considering detailed entropy generation rate of the cold-end systems. The results show that the entropy generation rate of the multi-pressure cold-end system is less than that of the single-pressure cold-end system when the total condenser area is constant. Moreover, the economic performance can be improved with the adoption of the multi-pressure cold-end system. When compared with the single-pressure cold-end system, the excess revenues gained by using dual- and quadruple-pressure cold-end systems are 575 and 580 k$/a, respectively.