Abstract: Tractors have been widely used as agricultural machinery in the field. This study aims to improve the overall economy of tractors using hydro mechanical continuously variable transmission (HMCVT). An economic control strategy was also developed to consider the transmission of the tractor. The engine and multi-stage HMCVT efficiency model were established using polynomial fitting and meshing power. The optimal economic control was also provided after optimization. A control strategy was established to balance the significant differences in the transmission efficiency among multiple stages of HMCVT with different gear ratios, taking the ratio of engine fuel consumption rate to HMCVT efficiency as the evaluation index. Simulated annealing (SA) and genetic algorithm (GA) were combined to utilize the natural evolution of GA, gradually improving the quality of economic parameter optimization; At the same time, the SA algorithm was used to avoid premature convergence with the increasing iteration times. Arithmetic crossover and non-uniform mutation were used to accurately obtain the most economical engine speed, torque, and transmission ratio. The new population G' was generated, when the GA globally optimized the initial population G. SA algorithm was then used to locally optimize the population G'. The Metropolis criterion was used to calculate the individual acceptance probability under isothermal conditions. Once the fitness function of the offspring population G' was better than G, G' was accepted as the optimal solution. A cooling operation was performed until the global optimal solution was obtained. The best overall economy of the machine was achieved by storing the optimal speed ratio, engine speed, and torque of the load in the control unit. The engine operating point and transmission speed ratio were regulated, according to the requirements of tractors. The combinatorial optimization was verified to compare the iterative performance of the SA algorithm and GA. Two algorithms fell into the local optima after 53 iterations, indicating their high accuracy of the combinatorial optimization. Finally, the control strategy of optimization was also validated to improve the overall fuel economy of the tractors. A vehicle model was constructed to simulate using the Simulink platform, compared with the traditional control strategy. The HMCVT speed ratio of the control strategy was verified using a test bench. The simulation results show that the control strategy was achieved in an average transmission efficiency of 0.93 for the tractor under plowing, which increased by 0.06 and 6.8%. The average fuel consumption rate was 279 g/(kW·h), which was reduced by 8 g/(kW·h) and 2.7%. The transmission efficiency of the tractor was also improved to save the fuel consumption. The HMCVT speed ratio was relatively low at the low speed and high loading. The tractor was output the sufficient torque after optimization. There was great consistency in the simulation and experiment. The control strategy effectively improved the overall economy of the tractor, while reasonably matching the working states of the engine and transmission. The findings can provide a theoretical basis for the engineering application and economic control strategy of HMCVT tractors.