The present invention discloses a wireless cloud computing system performance prediction method. The method comprises the following steps: confirming the mapping relation from a work flow to an AP, and calculating the related data; layering the task link of the work flow aiming at the condition that the users in the system are mainly blocked up at the communication process, and selecting a graph model with the maximum value of the longest path as a prediction model of the system performance of the cloud computing system reaching the maximum processing ability; about the condition that the users in the system are mainly blocked up at the computing process, selecting a graph model with the maximum value of the longest path as a prediction model; and finally, solving the average user processing time makespan of the system at different arrival rates through adoption of the mode of solving two functions to predicate the system performances. The performances of the computing system is predicated through the modularization method, and when the cloud computing system is huge and is difficult to realize in a short time, the wireless cloud computing system performance prediction method provides a frame method for prediction of the system performances and provides references for the problems such as the feasibility of the system and the parameter arrangement of the system, etc.