In the virtual machine placement problem, the traditional heuristic methods are not entirely applicable to the complex cloud computing environment, and the researches using intelligent algorithms lack the consideration of time overhead. To solve the above problems, a Memetic algorithm-based virtual machine placement (MAVMP) method is proposed. Firstly, The MAVMP method establishes a multi-objective optimization model for minimizing energy consumption, minimizing the service-level agreement violation times per active host (SLATAH) and maximizing resource utilization according to the operation situation of cloud data centers. Secondly, on the basis of resource requests, virtual machines are classified, improving the Memetic algorithm. Finally, the improved Memetic algorithm is used to solve the multi-objective optimization model, and then obtain the virtual machine placement plan. The results of simulation test show that the simulation data center using the MAVMP method to place virtual machines has good performances in energy consumption,resource utilization and service quality. Moreover, in contrast to the existing intelligent algorithm-based virtual machine placement method, the calculation time of the MAVMP method decreases sharply.