An iterative multiuser detection scheme with data estimate refinement is presented and analyzed. Iterative multiuser detection with data estimate refinement can be looked at as an application of the general turbo principle. FEC coding with modulation and CDMA spreading can be considered as a serially concatenated code. Turbo decoding is based on alternately decoding the two codes by the multiuser detector and the data estimate refiner. It is shown by simulations that iterative multiuser detection with data estimate refinement achieves a performance close to the matched filter bound in typical mobile radio systems like TD-CDMA. Thus, iterative multiuser detection with data estimate refinement helps to overcome the problems of low asymptotic efficiencies of linear multiuser detection like e.g. zero forcing estimation in TD-CDMA mobile radio systems with high system loads. Simultaneously the complexity of iterative multiuser detection with data estimate refinement is much smaller than that of optimum nonlinear multiuser detection.