Canonical correlation analysis is a multivariate technique, which is employed to examine relationships between two variable sets, each of that consist of two or more variables. In the analysis, by maximizing the relationships between two sets, it is aimed to obtain new variables called as canonical variates that formed by linear combinations of the original variables for each sets. In this study, it is aimed to examine the relationships between some fruit and kernel characteristics of Ankara region walnuts. Thus, the following eight traits Fruit Length (mm), Fruit Width (mm) Fruit Height (mm) Fruit Weight (g), Shell Thickness (mm), Kernel Weight (g), Kernel Ratio (%) and Filled-firm Kernel Ratio (%) of 364 walnut samples were measured. First five of these characteristics were grouped as X variable and the rest of as Y variable. As a result, the correlation between the first canonical variable pair was found as 0.979 (p < 0.01).