In recent decade, consensus problem in multi-agent systems has become an attractive topic for researchers and it has received great attention. Consensus problem has broad applications in many areas, such as robotics, unmanned air vehicle (UAV) formations, sensor networks, under water vehicles, etc. Consensus means a group of agents agree upon on their common states. Nonlinear multi-agent systems are more close to the real-world applications than linear multi-agent systems. Therefore, consensus problem in these systems is a significant topic for research. The main problem in nonlinear systems is how controller handle nonlinear dynamics. In previous researches many approximation based control techniques investigated for handling unknown nonlinear dynamics, such as adaptive control or neural networks. In this thesis, control strategies based on adaptive fuzzy wavelet networks (AFWNs) are presented to compensate nonlinear dynamics and solve consensus problem. AFWNs can approximate a broad range of nonlinear functions with fast learning ability and desired accuracy. Due to the approximation capability of AFWNs, this thesis uses AFWNs to compensate nonlinear dynamics. In the first part of this thesis, first order nonlinear multi-agent systems are studied in two cases. First, consensus in leaderless systems and then consensus in leaderless systems with unknown state time delays. Uniformly ultimately bounded stability of the control algorithm proofed and proper adaptive laws for estimating AFWN parameters are obtained. In the second part of this thesis second order nonlinear multi-agent systems are studied in three cases. First, leaderless systems then, leader following systems and finally leader following systems with unknown state time delays. Uniformly ultimately bounded stability of the control algorithms using proper Lyapanov functions are proofed and proper adaptive laws for estimating AFWN parameters are obtained. Key Words : Nonlinear Multi-Agent systems, Consensus, Fuzzy Wavelet Networks, Time Delay.