Three 36-year-old beams are taken as research objects to estimate the fluctuation scale θ of structural dimensions and material property parameters of reinforced concrete structure in service. First, the protective layer thickness C_d and the concrete compressive strength f_c of service bridge component are measured, and the ordinary kriging method is used to interpolate the measured data and gather effective data. Second, the θ values of C_d and f_c are estimated by two methods, namely autocorrelation function (ACF) and semivariogram function (SVF). Thereafter, the spatial distribution of R of each corroded steel bar is studied on the basis of measured values of the pitting factor R at different positions of 246 corroded steel bars. On the basis of one-dimensional random field theory, each beam is discretized into several small elements. According to the fluctuation scale obtained by different estimation methods, the random variables are transformed into spatial correlation variables, and the parameters of each discrete element are obtained. Then the bending capacity of each discrete beam element and the corresponding whole beam are obtained on the basis of the value of parameters of each discrete element. Finally, the two estimation methods are compared and analyzed by comparing the theoretical moment value and the bearing capacity test results of the corroded beams. Results show that the mean values of θ of f_c and C_d obtained by the SVF method are 2.242 and 2.467, respectively, which are larger than that obtained by the ACF method. Moreover, the fluctuation intensity of concrete compressive strength is smaller with an increase of θ value, and no firm correlation is identified between the spatial distributions of each pit on the surface of a corroded steel reinforcement. The bending moments of three beams from the SVF method are in better agreement with the experimental values than those predicted from the ACF method. In addition, the average relative error of the theoretical bending moment value of all beams based on the SVF estimation is 6.52%, which is 33.19% lower than that obtained by ACF method. Therefore, the SVF method is more effective in describing the spatial properties of material and geometrical parameters compared with the ACF method.