Conventional MultiUser Detection (MUD) algorithms for the Multi-Input, Multi-Output-Orthogonal Frequency Division Multiplexing (MIMO-OFDM) system fail to consider the detection performance and algorithm complexity simultaneously. To address this problem, a new Joint Intelligent MUD (JI-MUD) algorithm that aims to improve the MUD performance of the MIMO-OFDM system and to reduce the complexity of the algorithm was proposed. First, a new MUD method based on the genetic algorithm (GA) for the MIMO-OFDM system was introduced. Utilizing the results of the Minimum Mean Square Error (MMSE) algorithm as the initial population and the criterion of the Maximum Likelihood (ML) algorithm as the fitness function, the proposed algorithm performs genetic operations through the roulette wheel selection operator, two-point crossover operator, and adjacent bit reverse mutation operator, which generate a new population. Second, a Hybrid GA (HGA) was presented by combining the simulated annealing and particle swarm optimization algorithms. The extended study on the HGA complexity and performance was conducted from a mathematical perspective. Finally, a quantitative analysis on the complexity and convergence of the HGA, as well as the correlation of the fitness function, was implemented. Research results demonstrated that the convergence measurement function value of the proposed HGA is 0. Furthermore, its signal-to-noise ratio is approximately 1 dB lower than that of the MMSE-MUD algorithm and approximately 1 dB higher than that of the ML-MUD algorithm when the error rate is 5 × 10 ˗2. This finding indicates that the proposed algorithm perform better than the MMSE algorithm and approach the ML algorithm at the cost of appropriate complexity. These research results can better balance complexity and performance and guide the follow-up on the MUD development in the MIMO-OFDM system.