In this paper, a depth cue method based on blur effect in augmented reality is proposed. Different from previous approaches, the proposed method offers an algorithm which estimates the blur effect in the whole scene based on the spatial information in the real world and the intrinsic parameters of the camera. We implemented a prototype using the proposed method and conducted two user tests on how the users might perceive the blur effect rendered by different blurring methods. The test settings are introduced and the results are discussed. The test results show that our blur estimation method is acceptable for moving virtual objects. We also find that the users might prefer a stronger contrast of blur than the blur consistent to the background.