To deal with the complex constituent of measured stress and low efficiency of utilizing the vast measured data, this paper developed a new approach for the prediction of stress development and condition assessment of bridge structures based on statistics theory. This method separates the stress with seasonal distribution and stochastic distribution from the whole measured stress using the statistics theory. By doing this, the stress development due to the structure deterioration can be observed and theoretically predicted, which gives a quantifiable structure assessment. Every data measured in the realtime measurement can be used to give the assessment results and each newlymeasured data in future can be applied to further update the results. A case study was finally conducted to show and prove the advantages of the proposed method.