In Korea the importance of management of bridges has been recognized over a couple of decades, resulting in the development of database and various bridge management assistant tools by both government and private sectors. However, none of them has truly included the expected features of the Bridge Management System (BMS) for the next generation such as the quantification of the effect of maintenance interventions. As a result, life-cycle cost analysis has been performed within the limits of analyses including uncertainties in terms of application time of maintenance interventions and of overall costs for bridge maintenance only. Recognizing the problem, a new research project has been launched to construct a bridge management system which has enough flexibility and extendibility for the next generation BMS. This system has been constructed based on new theoretical background and included many features which were not included in the conventional BMS. The system has been developed for steel bridges first. In this paper, background and important aspects of the novel management system developed in Korea for steel bridges is represented. The fundamental difference between management systems is due to the difference of assessment system. Moreover the difference of assessment is often originated from the difference in indexing system. Fig. 1. shows the characteristics of different assessment indexing systems. Each system has its own advantages and disadvantages. In the developed system, multi-indexing system has been implemented to gather available information from all kind of sources. Quantification of the effect of degradation of members is a difficult and tedious process because of the diversities of bridges depending on designer's selections and loading condition. Therefore, it is very important to develop a simple but reliable enough method to compute the lifetime performance of the target structure. Solution may lie on a simple regression model established based on enormous analysis results obtained by considering expected possibilities of loading conditions and structural properties. The Response Surface Method (RSM) has been proposed and used successively to (Graph Presented), construct the performance profile of deteriorating bridge members. Among 1,954 steel box bridges in Korea, 633 steel box bridges maintained by the Ministry of Construction and Transportation has been analyzed to decide the design variables. Based on the results, bridge types have been classified depending on length of bridges, number of spans, span length, width of bridge, different combinations of different span lengths, number of lanes, number of steel boxed, height of steel boxes, variation in thickness of steel plate, support conditions and so on. Dependency between design variables was investigated to reduce the number of unimportant design variables. Lots of dependent variables could be eliminated and important design variables are identified. For instance, it is concluded that only three design variables, the length of bridge, thickness of concrete slab, and the width of girders are dominant to the behavior of simply supported one-span steel box girder bridges. For two-spans bridges two variables, the length of the first span and the web thickness at the middle of span are added to create the response surface models. Time dependent performance of the three-spans bridges can be analyzed by using the similar variables as of the two-spans bridges except that the length of the first span is replaced by the ratio of the length of the second span to the total length of bridges. It is denoted as explicit model because the explicit performance profile is obtained from numerical analyses. Quantification of degradation (and also quantification of maintenance effect) based on the direct analysis and the response surface models are not always possible to be established. To mitigate this problem, two additional methods have been applied. One is to use the historical record storing in KOBMS, a database which has been reformed from the BMS accumulating data since 1989. Another method is to use experts' opinions even though this method has to deal with subjective information. In this case, the performance profile is assumed as a linear model and denoted as the implicit model. Quantification of maintenance effect is basically the same to the problem of quantification of degradation. For maintenance interventions without any direct relationship to the structural performance, such as cleaning drainage, are computed based on the frequencies of occurrences based on historical records. The other maintenance interventions causing variation in structural performance, such as essential maintenance interventions, are analyzed and their effect on structural performance is quantified by the response surface model. Number of design variables to construct the response surface model changes depending on the characteristics of maintenance interventions. For instance, two more variables, time of application and average daily truck traffic volume (ADTT) have added to the initial design variables to compute response surface equations. There are many elements to be considered to make an efficient but comprehensive bridge man-agement system. Among them the items which will be discussed in the following sections have been being considered extensively. Eventually, cutting edge technologies combined with advanced assessment and analysis methods will produce the smart structural management system for the next generation.