The problem of simultaneous fault detection and consensus control (SFDCC) of linear continuous-time multi-agent systems is addressed in this paper. A mixed H-infinity/H- formulation of the SFDCC problem is presented and distributed detection filters are designed using only relative output information among the agents. With our proposed methodology, all agents reach either a state consensus or a model reference consensus while simultaneously collaborate with one another to detect the occurrence of faults in the team. Indeed, each agent not only can detect its own fault but also is capable of detecting its neighbor's faults. It is shown that through a decomposition approach the computational complexity of solving the distributed problem is significantly reduced as compared to an optimal centralized solution. The extended linear matrix inequalities (LMIs) are used to reduce the conservativeness of the SFDCC results by introducing additional matrix variables to eliminate the couplings of Lyapunov matrices with the system matrices. It is shown that under a special condition on the network topology the faulty agent can be isolated in the team. Simulation results corresponding to a team of autonomous unmanned underwater vehicles (AUVs) demonstrate and illustrate the effectiveness and capabilities of our proposed design methodology. (C) 2016 Elsevier Ltd. All rights reserved.