Abstract: The bridge weigh-in-motion (B-WIM) technique could derive the critical information of a vehicle like axial weight and wheelbase from bridge responses. Compared with pavement-based WIM technique, B-WIM's installation and maintenance process could avoid the impact on normal traffic, which made it a convenient technique for bridge upper load monitoring. However, there are still some issues such that closely distributed group axles cannot be identified, and that sensors' durability is inadequate. Concentrating on these issues and utilizing the macro-strain time history captured from long-gauge FBG sequence, a critical index: macro-strain curvature was proposed, which is directly related to a vehicle's information. Based on this index, a new B-WIM method was established to simultaneously identify a vehicle's speed, wheelbase and axial weight. Then on the grounds of an experiment verified vehicle-bridge coupling simulation, a bridge's macro-strain response was simulated under various parameter scenarios. This B-WIM method's measurement accuracy was analyzed under different vehicle types, speeds, road roughness degrees, sensors locations and gauge lengths. The results showed that: this method's overall measurement accuracy on vehicle and on wheelbase is rather good. The influence of each parameter is minor and relevant error is less than 1%. In contrast, the results of axial weight are not ideal, especially for axles with small wheelbase or group axles, but its accuracy can be dramatically refined by reducing the sensor's gauge length and the problem that closely distributed group axles are hard to be identified would also be solved. Meanwhile, this method's precision would not be impacted under different sensor locations, which would effectively increase the convenience in system installation and maintenance.