Civil engineering structures, especially bridges, are mostly situated in adverse environmental conditions. More specifically steel bridges are very much prone to damage due to age and other unfavorable environmental conditions. Early detection of damage is key to preventing failures of bridges. In practice, these distressed structures are inspected by experienced engineers to determine possible damage state. The detection of structural damage and identification of damaged elements in a large complex structure (like a bridge) is a challenging task. Artificial neural networks have gained popularity over conventional methods of damage identification due to their pattern recognition and multi-class discrimination capabilities. Further, recently non-contact measurement techniques, typically utilizing light or lasers, have emerged, offering promising applications in static deflection measurement. In this regard, the development of a neural network for damage identification of a warren-type truss bridge using static test data is taken as the main objective of the present study. Damage is identified by classifying the deflection pattern evolved due to the occurrence of any damage within the truss. For this purpose, a network considering data from three different trusses has been built. The network is built with two different approaches using the same data set, an ordinary feed-forward network and a modular network. The developed network shows adequate generalization capability. The proposed approach is useful in quick assessments of aging steel truss bridges which is needed for decision-making about bridge maintenance and repairs.