Abstract: Micro-Genetic Algorithm（μGA） is a new GA technique based on the use of small population and some very small genetic parameters.It considerably reduces the number of operations as well as the total cost of the proposed detector.A discussion on the population of μGA has been made.An improved algorithm that employs μGA for the Multi-User Detection problem（μGA MUD） is proposed.Numerical results have shown that in aspect of bit-error rate and “near-far” resistance，the μGA MUD is close to the Optimum Multi-user Detector（OMD）.And in aspect of computing complexity，the μGA MUD is much better than OMD.