In this doctoral dissertation, a new proposal was developed for the flexural bucklingresistance of welded I-section stainless steel columns exposed to fire considering theinfluence of loading history. A finite element procedure has been developed fordetermining the flexural buckling resistance in fire, taking into account the influenceof the loading history. The finite element procedure was coded in the Pythonprogramming language while using the Abaqus solver. The finite element proceduresimulates the real behavior, where the column is first loaded and then exposed to thefire action according to a standard fire. The numerical procedure was verified inrelation to experimental tests on welded I section stainless steel columns in fire fromthe literature. Using the developed and validated numerical procedure for determiningthe ultimate load capacity, the shortcomings of the rules for the buckling resistanceload capacity according to the rules given in EN 1993-1-2 were confirmed.The poor accuracy and large dispersion of results of prediction of the flexuralbuckling resistance of stainless steel columns in fire according to EN 1993-1-2, butalso according to recent proposals, indicated that the expressions for determining the buckling resistance in fire in the existing form cannot adequately predict resistance. New contributing parameters must be included on the expression side in order to improve the accuracy of the prediction. Two main contributing parameters, the ratioof flanges and web areas ⁄ and the ratio of flange and web thicknesses ⁄of I sections, were analyzed in order to indirectly include the influence of loadinghistory on the side of the expression for determining the buckling resistance. A new set of rules for determining the buckling resistance of axially compressedstainless steel columns under the action of fire is proposed. As a reference strength,the strength corresponding to a 2% plastic strain , was used, which is in accordancewith the rules for carbon steel columns in EN 1993-1-2. The expression for thereduction factor is calibrated against numerically determined buckling curves definedon the basis of the results of parametric numerical analyses. Parametric numericalanalyzes covered a wide range of relative slendernesses ̅, a range of criticaltemperatures of practical importance, compact and slender cross-sections, a widerange of cross-section dimensions and three different types of stainless steels,austenitic, duplex and ferritic. The coding of the numerical methodology for thebuckling resistance assessment with the complete calibration process of the newlyproposed expressions for determining the buckling resistance in the Pythonprogramming language, enabled the realization of a large number of numericalanalyzes and their use in the calibration process (a total of 6000 numerical simulationswere carried out, of which 3000 were used in the calibration process).The accuracy of the new proposal was evaluated against the basic statisticalparameters, while the reliability of the new proposal was evaluated against thereliability criteria for the design of steel elements in fire set by Kruppa. A comparativeanalysis of the accuracy of the new proposal, recent proposals from the literature andthe proposal according to EN 1993-1-2 showed that the new proposal provides a farmore accurate prediction of the buckling resistance compared to the remaining twoproposals, which enables a more economical and safer application of these elements.