In view of the negative influence of rotor wake (i.e. the aero-olfactory effect, the short name is AO effect) on gas distribution, this paper proposed an AO-effect elimination algorithm for rotor UAV (Unmanned Aerial Vehicle) based three-dimensional (3D) gas distribution mapping (GDM). Firstly, two types of AO effects, i.e. the lateral effect and the vertical effect, were summarized based on experimental phenomena. Secondly, the wake induced gas-patch path was calculated based on the prescribed wake model, and the AO effect was eliminated by estimating the original position of the detected gas patch through backstepping along the induced path. Thirdly, considering the different diffusion speeds along different directions, an ellipsoidal Gaussian distribution rather than the traditional spherical Gaussian distribution mapping method was put forward, which was validated via an experiment conducted using four wireless sensor nodes. The proposed AO-effect elimination algorithm combined with the ellipsoidal kernel GDM has been tested using a robot active olfaction simulator, and the results show that it is more effective in characterizing the gas distribution.