Two functionalization technichques are applied to the fatigue data base edited by the Society of Materials Science, Japan. Among many data, 273 sets of S-N curves of rotary bending test data of steels in S-N forms, were chosen for functionalization, expecting better usage of a data base on three points; i) to estimate a S-N curve of a material whose test conditions or chemical composition is not included in the data base, ii) to express the reliability of data in terms of standard deviation, iii) to detect miss-inputted data. A S-N curve is expressed in three functions with 16 variables (twelve chemical components, radius of specimen, stress concentration factor, testing frequency and tensile strength). One of the functionalization technique is a statistical binary regression analysis, and the other is an interpolating technique in a higher dimensional space.The standard deviation of data points from the regression funtion was bigger than that from the interpolated function. The regression function showed divergent tendencies within the range of analysis. Some examples of the interpolated function projected to lower dimensional spaces were illustrated to show reasonable results. The shape of interpolated curved surface is analyzed at a structural steel point, to detect effective or non-effective factors to shift the S-N curve among 16 variables. The interpolation showed a good technique to functionalize a data base.