The cross-section profile of the rolled strip is characterized by geometrical parameters such as wedge, convex, difference of thickness, edge wedge and others. Techniques for calculating the values of these parameters are known and generally accepted. However, there are features of the cross-section profile of rolled strips, such as local thickenings/thinnings, the methods of calculating values of which are not common: practically every scientific school of specialists of rolling production use their own techniques, which often produce different results for the same cross-section profiles. The problem of identifying and calculating the local thickenings/thinnings parameters of the rolled strips cross-section profile is to define a so-called “zero level”, the excess/understatement of which is a sign of local thickenings/thinnings. The paper analyzes the accuracy and adequacy of the calculation of the cross-section profile parameters of rolled strips for local thickenings/thinnings. A new method based on statistical methods is proposed. The target function that the thickness distribution across the width of the rolled strip must correspond to is a symmetrical quadratic parabola. However, the actual distribution is always different from the target one for a number of reasons, such as ring wear of the work rolls. In the first step, in the proposed technique, the Walter-Shewhart procedure (control cards) eliminates as emissions of strip thickness values that are dramatically different from the target distribution. But since without excluding the nonlinear (parabolic) component of the measured cross-section profile this procedure cannot be applied, it applies to the first derivative of the cross-section profile thickness distribution function. The result of the developed method is a “zero level” according to which the local thickening/thinning parameters are calculated.