This paper studies the tracking problem for a class of nonlinearly parameterized multi-agent systems with the powers being unknown time-varying functions. The essence of the multi-agent systems and the serious uncertainties distinguish the question in this paper from the existing works. By combing the algebraic graph theory, adding a power integrator method and adaptive technique, a new distributed adaptive tracking controller is presented with a dynamic estimating the unknown parameters, which guarantees that both the system state and the control input are globally bounded and meanwhile, the tracking error can be adjusted to arbitrarily small. Finally, we offer a simulation to interpret the validity of the control scheme design.