In the Internet-of-Things scenarios, unmanned aerial vehicle (UAV), as a popular aerial platform, is calling for ever-increasing computing support. This letter proposes a novel mobile edge computing (MEC) framework for UAV with the assistance of the reconfigurable intelligence surface (RIS), where a UAV offloads the computation tasks to ground access points (APs) with the assistance of an RIS, during which non-orthogonal multiple access (NOMA) scheme is employed. We maximize the UAV’s computation capacity by jointly optimizing the reflecting phase shift, communication and computation (2C) resource allocation, decoding order, and UAV’s deployment. Specifically, we first derive the reflecting phase shift by invoking the concave-convex procedure (CCCP) method and the semidefinite relaxation technique. Next, we obtain the 2C resource allocation by using the CCCP method. The decoding order and the UAV’s deployment are finally solved via proposing a grid search (GS) method. Numerical results demonstrate that: 1) the computation capacity is greatly improved by the design of RIS; 2) NOMA scheme outperforms orthogonal multiple access scheme; 3) the proposed GS method achieves significant performance gains, as compared with the traditional convex approximation method.