This paper presents a method to identify the parameters of continuous and discrete MIMO Takagi-Sugeno (TS) fuzzy models. The proposed method combines the Levenberg/Marquadt (LM) optimisation algorithm and least squares (LS) method. The LM algorithm is used to estimate the Gaussian membership function parameters whereas the parameters of linear local models is determined using a LS algorithm. To illustrate the effectiveness of the algorithm, an application to a laboratory quadruple-tank process is presented.