This paper considers the fully distributed tracking consensus problem for general nonlinear multi-agent systems with a leader whose control input is nonzero and bounded. First, a new class of distributed state observer for the leader is proposed without the knowledge of the upper bound of the leader's input. Then, the situations that followers are affected by disturbances with unknown upper bound or disturbances generated by exosystems are investigated. Specifically, two distributed control protocols based on the distributed state observer, neural networks, and adaptive laws are proposed. Finally, simulation examples are provided to illustrate the theoretical results. Index termstracking consensus, general nonlinearity, external disturbances, unknown upper bound.