The invention provides a self-adaptive fuzzy underwater vehicle control method realized through the genetic algorithm-based optimizing process. Self-adaptive parameters are used as the alleles of chromosomes and the population initialization is carried out. Any individual in an initial population is selected to be substituted into an adaptive fuzzy controller to control a given desired value. After the control is finished, the integral performance index and the fitness function value corresponding to the selected individual are calculated. After that, another individual is selected to be substituted into the controller for control purpose. In this way, the individuals in the current population are substituted into the controller one by one for control purpose, and then the fitness values of all the individuals are obtained. After that, the convergence judgment is carried out. If the condition is met, the iteration is stopped. An individual with a highest fitness value is outputted as an optimal solution. If the condition is not met, the selection, crossover and mutation operation is conducted on the current population and then a next population is generated. After that, the cyclical iteration is continued until the convergence condition is met. The obtained set of adaptive parameters is applied to the actual motion control of an underwater vehicle and an optimal control effect can be obtained.