Bridge testing provides a much needed means of correctly assessing the strength or load carrying capacity of a bridge. When carefully planned and executed by employing properly selected equipment and complemented by efficient identification and confirmatory analysis, it is obviously the surest. way of determining the load carring capacity of the bridge. The lack of as-built construction plans and original design calculations poses difficulties in determining the magnitude of the live-load stresses in the structural members using typical analytical models. Thus for bridge engineers a major concern is the lack of rational procedure for reliable analytical modelling. A reliable analytical model should simulate dynamic characteristics, all ofthe important resistance mechanisms, the effects of any existing damage and deterioration on these mechanisms. The field measurements taken during bridge test scan also be used to modify analytical model developed depending on its as-built construction plans and original design calculations, to produce the final reliable field-testing-based analytical model. Since the model has been based on actual field measurements, and assuming that the structure behaves in a linear fashion, relatively accurate stress-level estimation for higher loads (such as rating or permit load) can bemade. The developed model can also be used to estmate the remaining fatigue life ofthe bridge.Once a bridge has been tested and a model developed, it can saved in the computer and compared with test run at periodic intervals. In this way , bridge engineers may then be able to determine whether a bridge's load capacity is changing as a functionof time or not. This model can also be used to develop retrofitting schemes. For this reason, it is important. to have an analytical model that can accurately predict the static behavior of the bridge structure.In this proposed study Automatic Remodelling Technique in which a systematic optimization scheme is utiIized to adjust the structural properties in an automated fashion to refine the initial computer model to improve the correlation to the measured strains, is developed.Refined computer model estimation is based on measured strain data induced by semi-static loads such as a slowly moving test locomotive on a railway bridge.There is limited work done using static test data. In the literature there are studies,which use parameter estimation methods using both static and modal data. Simulated measurements of static deflections or vibration modes were successfully used for parameter estimation. There is another method of parameter identification using static test data measured at a subset of degrees of freedom (DOF). Structural stiffness were successfully estimated at the element level, including large changes and even element failures for stable structure. This method was limited to force and displacement measurements at the same DOF .Incomplete sets of applied static forces and displacements are also used to estimate element stiffness. In this method, the unmeasured displacements were also assumedto be unknoW and estimated. On one hand, this assumption simplified the error function to be minimized; on the other hand, the number of unknows was increased to the unknowll parameters plus the unmeasured displacements. Another parameter identification technique locates and characterizes loose joints of a deployable spacetruss using actuator-induced static loading and unloading.Another important method for parameter estimation of linear-elastic structures usesstatic strain measurements was demonstrated only for two dimensional truss andframe type structures. Methods for identifying the axial stiffness properties of a truss structure, only is also available.Above cited studies do not incorporate additional stiffness sources such as additional stiffness effects of gusset plates, joint looseness of the riveted truss joints and lockedsliding support effects.Thus, in this study an efficient field testing based model updating algorithm which was a 3 dimensional Finite Element Program, based on Direct Element Method of space rigid frame analysis is written and developed.In this finite element program rigid bars are considered at two ends of each elastic prismatic elements to simulate the gusset plate characteristics, semi rigid connection sand irregularities of the bridge structure (see Figure 1 ). The model has four vertical,and two horizontal spring elements at support points to simulate the effective vertical stiffness of the combined bearing and abutment structure, and the longitudinal restraints at the sliding bearings . In this proposed study finite element algorithm based on the Direct Element Methodof space rigid frame analysis is developed and used in order to compute the joint displacements and the internal forces of the members.The static finite element equation for a structural system defined as, ASAT.X=p (1) where X gives the column matrix for the joint displacements and p represents the--matrix of external joint forces acting at the joints.In the joints of a space rigid frame, the details of rigid beam to column connectionsare usually idealised, but in the case of large gusset plates in truss bridge the endjoints of flexural element act as a definitely rigid bars Since the riveted connection of the bridge truss members can not be considered fully fıxed and since very often the rigidity contributions of gusset plate to the truss members may take place, therefore the effective axial and flextural stiffness of a typical member of any steel riveted truss bridge are expressed as a function of it scross sectional area, cross sectional moment of inertia and the rigid bars at the bothends to simulate the gusset plate and semi-rigid connection as is seen in Figure 2.In this Figure F stands for the cross sectional area ofthe member, Ix,ly are moment of inertia about two transverse axis, E and L stands for elasticity modulus and element length, ar and br are rigid bar length at two ends i and j .Properties such as torsional rigidity of the element do not significantly affect the models and therefore they were excluded from the optimization by fixing their values. Except E and L other stiffness parameters are considered as optimization variables and adjusted automatically in the developed program along with the spring constants of support elements considered in the initial model. The optimization variable vector is defined as:x= {Fı,lxl,Iyl,arl,brı,...,Fn,Ixn,Iyn,abr,brn} (2)Depending on the obtained values of the parameters the structure is reanalyzed untila high correlation to the measured strains is reached.The optimization algorithm proposed herein attempts to minimize the objective function considered as the sum of squares of the differences between the experimentally and analytically obtained strains for appropriate strain transducer location.