Recent advancements in both hardware and software for mobile devices have allowed developers to create better mobile Augmented Reality (AR) experiences, which has led to an increase in the number of mobile AR applications and users engaging in these experiences. However, despite a broad range of mobile AR applications available to date, the majority of these applications that we surveyed still primarily use surface gestures, i.e., gesturing on the touch screen surface of the device, as the default interaction method and do not utilise the affordance of three-dimensional user interaction that AR interfaces support. In this research, we have investigated and compared two methods of gesture interaction for mobile AR applications: Surface Gestures, which are commonly used in mainstream applications, and Motion Gestures, which take advantage of the spatial information of the mobile device. Our goal is to determine if motion gestures are comparable or even superior to surface gestures for mobile AR applications. To achieve this, we have conducted two user studies: an elicitation study 15 and a validation study. The first study recruited twenty-one participants and elicited two sets of 16 gestures, surface and mobile gestures, for twelve everyday mobile AR tasks. This yielded a total 17 of five hundred and four gestures. The two sets of gestures were classified and compared in terms of goodness, ease of use, and engagement. As expected, the participants’ elicited surface gestures are familiar and easy to use, while motion gestures were found more engaging. Using design patterns derived from the elicited motion gestures, we proposed a novel interaction technique called ”TMR” (Touch-Move-Release). We developed a mobile AR game similar to Pokemon GO to validate this new technique and implemented a selected gesture chosen from ´ the two gesture sets. A validation study was conducted with ten participants, and we found that the motion gesture enhanced engagement and provided a better game experience. In contrast, the surface gesture provided higher precision resulting in higher accuracy and was easier to use. Finally, we discuss the implications of our findings and give our design recommendations for using the elicited gestures.