A method based on crisscross optimization algorithm (CSO) is introduced to solve combined heat and power economic dispatch (CHPED) problem. CSO adopts a double-crisscross search mechanism, introducing expand factor to increase global search ability in horizontal crisscross, introducing the concept of dimension-crisscross in vertical crisscross, to avoid falling into dimension local optima. The CSO implements global parallel search, which can effectively accelerate the convergence rate. The proposed method is illustrated for a 48-units system that contains power-only units, cogeneration units and heat-only units, with which the model of CHPED problem is established. The simulation results show that the proposed algorithm is feasible and effective in solving the CHPED problem, providing a better method for the actual scheduling system.