Large amplitude oscillatory shear (LAOS) behavior of complex fluids, which form microstructures depending on their deformation history, has been investigated by using a network model. According to recent experimental observations, the LAOS behavior of complex fluids could be classified by at least four types: type I, strain thinning (G', G" decreasing); type II, strain hardening (G, G" increasing); type III, weak strain overshoot (G' decreasing, G" increasing followed by decreasing); type IV, strong strain overshoot (G', G" increasing followed by decreasing). To understand such complex behavior, we have applied a general network model. As there is little information available on the form of creation and loss rates of network junctions, we have modeled the creation and loss rates as exponential functions of shear stress. By adjusting the model parameters that define the creation and loss rates, the types of LAOS behavior observed in the experiments could be reproduced. Despite highly simplistic modeling, the model reproduced the types of LAOS behavior observed in the experiments, which means that the behavior can be explained in terms of the model parameters, that is, the creation and loss rates of network junctions. It is also suggested that the LAOS behavior can be effectively used as a tool for classifying complex fluids. (C) 2003 Elsevier B.V. All rights reserved.