Detecting multiple simultaneous line outages in power transmission networks is known to be a challenging problem due to the number of hypotheses that grows exponentially with the network size. A low complexity message passing algorithm is proposed for multi-line outage identification, which exploits the underlying sparse structure of the network topology in power systems. First, a factor graph is established that characterizes jointly the power system and the sensor network monitoring it. For inferring line status, the mixed integer and continuous variables and the loopy structure of the factor graph make it difficult to use conventional message passing algorithms. Exploiting the power flow equations, efficient message representation and new techniques in message passing algorithms are developed. Simulation results demonstrate that the developed algorithm can effectively identify an arbitrary number of simultaneous line outages in real time.