This paper presents a distributed location method for indoor gas source based on extended set-membership filtering. Compared with the statistical estimation method based on stochastic model (extened Kalman filter and unscented Kalman filter), this kind of method only needs to know the boundary of the unknown noise, regardless of the randomness of the noise. The static turbulence model is utilized to calculate iteratively the positioning error boundary, and the actual value of the position state is effectively included in the estimation range, so that the positioning credibility can be achieved. Meanwhile, the least squares method is introduced to carry out the initial positioning to overcome the problem of initial point selection of the extended set membership filtering. Finally, indoor positioning simulation experiment based on the wireless electronic nose is conducted and the feasibility and validity of the proposed method are verified.