A mechanical arm motion control method comprises the following steps of: (1) establishing complete corresponding transformation transfer matrices of n connecting rods, working out linear velocities and angular velocities of the connecting rods, (2) calculating centers of mass of the connecting rods to obtain inertia tensors, working out total kinetic energy and total potential energy of a mechanical arm, obtaining a partial derivative of the difference between the total kinetic energy and the total potential energy to a generalized coordinate q of the mechanical arm, obtaining a relationship between a driving force vector tau and q, q', establishing a mathematical model, and (3) performing iterative calculation on set time with a variable gain iterative control algorithm for each time point, acquiring an error between each time point and expected pose qd, subtracting the errors from the current pose q(0) to obtain an initial error e'1(0), taking the driving force vector tau as system input, performing the iterative calculation on the set time with the variable gain iterative control algorithm, obtaining the driving force vector tau required to control the mechanical arm to reach the expected pose at each time point. The method is higher in control precision and good in flexibility, and is applicable to non-linear and time-varying occasions.