A damage detection procedure using dynamic long-gauges strains in the modal space has been proposed to process the ambient vibration data of a steel girder bridge.The theory basis of the proposed algorithm is as follows:(1)It is derived that the peaks of a column of long-gauge strain frequency response functions(FRFs)in each mode make the strain mode shapes;(2)It is proved that the magnitude ratios between the macro strain FRFs estimated from ambient vibrations and the analytical ones are mode-dependent,thus it provided a way to estimated strain mode shapes from ambient vibration data;(3)A simple but effective way is proposed based on the above theoretical findings for structural damage detection from mass dynamic strain data.Application of the proposed method is performed to a steel girder bridge using field test data for illustration purpose.