Estimation of the pairwise correlation coefficient between random variables for situations where no joint sample of values is available will usually be a difficult task. However, in some cases it is possible to identify underlying physical mechanisms represented by one or several more fundamental variables which influence all the “output” variables under consideration. In cases, available data sets for the underlying basic variables and input-output regression relationships can be utilized for the purpose of quantifying the correlation coefficients. An example of application related to estimation of characteristic ice parameters and their correlation along the Northern Sea Route (NSR) is given. For design and operation of ice-capable vessels in the Arctic, ice conditions should be considered as key factors since they are closely related to the ship ice loads and structural response. In this work, four representative sites are selected in order to study the relevant ice condition parameters. Specifically, ice thickness is estimated by applying the accumulated freezing degree days (FDD) method by utilization of available air temperature records. The corresponding ice flexural strength is subsequently calculated according to formulas which are given in the literature based on previous studies. The synthetic joint sample of ice thickness and flexural strength is then applied for the purpose of estimating the correlation coefficient between these two random variables.