Microbial inactivation and chemical conversion in water treatment reactors depend on the degree of flow segregation and earliness of mixing (i.e., micromixing).However, little is known about micromixing in full-scale water treatment reactors. This study used the seasonal conversion of residual disinfectant between chloramines and free chlorine as reactive tracers to evaluate micromixing in full-scale baffled and unbaffled clearwells. Effluent tracer concentrations were modeled using segregated flow (SF), maximum mixedness (MM), tanks-in-series (TIS), and reactor network (RN) models. Breakpoint reactions were most accurately predicted by the TIS model for both clearwells. The MM model was only accurate in the unbaffled clearwell under steady flow conditions. Segregated flows with different residence times (e.g., SF, some RNs) poorly represented observations. Micromixing was significant in the full-scale reactors studied, and the TIS model most accurately represented observed micromixing. When modeling multiple reactions or nonfirst-order reactions, reactor models should incorporate micromixing.