Abstract: With respect to the premature convergence and slow convergence rate of Adaptive Genetic Algorithms Multiuser Detector（AGA-MUD），applied the nonlinear Sigmoid function to the adaptive genetic algorithms and combined the resisted multiple-access interference performance of decorrelation algorithms，a new Multi-User Detector based on Improved Adaptive Genetic Algorithms and Decorrelation algorithm（IAGAD-MUD） is proposed.Simulation results show that IAGAD-MUD is better than AGA-MUD and Simple Genetic Algorithms Multi-User Detector（SGA-MUD） not only in resisting multiple-access interference and near-far effects，but also in capacity of system.