Most techniques used to adjust climatological time series for inhomogeneities require a homogeneous reference series for comparison with each time series being evaluated. However, creating homogeneous reference series from data with unknown inhomogeneities presents many obstacles. Although a truly homogeneous reference series may, in fact, not be obtainable, in this article we present a method that has proven successful at minimizing inhomogeneities in the creation of reference climatological time series. The method utilizes a multistep process based on non-parametric statistics and an alternative way of relating two time series through correlation analysis. This alternative uses the year-to-year change in the variable under consideration (e.g. temperature) as the basis for correlation.