A new method for bridge structural health monitoring using a wireless sensor network is introduced. This method uses heterogeneous sensors connected in a wireless sensor network and aims to detect damage in the structure at early stages while reducing the dependence of the method on exact knowledge of the excitation of the structure.  The wireless network adds many constraints in the algorithm used, and this work focuses on low energy consumption, low computational power and wireless sensors which may have unsynchronized clocks. The network architecture has two layers: the micro layer for local damage detection and the macro layer for data fusion.  At each node, two related measurements, acceleration and strain, are compared using appropriate models of the healthy bridge and residuals. If the measurements are different than expected, the structure may be damaged. At the higher level, all responses from the nodes are combined to detect whether the structure is healthy or not. Novel contribution of this work include the problem formulation using heterogeneous sensors and a two-tiered network architecture, the detection scheme at each node using weighted thresholds and the data fusion at the higher level, and the associated simulations based on a finite element model of a cantilever beam. The goal has been to combine ideas from many fields in order to set the foundation for an approach in damage-detection using wireless sensors networks.