A statistical test is proposed to examine the relationship between the mean and variance of the relative length of Y chromosome (as measured by Y/A2 or Y/F indices) in various racial groups. It is shown that the variance of Y chromosome length within a population is quadratically related with its mean if the Y chromosome length variation is caused only by variations in the number of copies of repeated DNA sequences. No such relationship is observed in the available data on relative lengths of the Y chromosome in various racial groups. We thus conclude that the length polymorphism in human Y chromosome may not be solely due to variations in the amount of Y chromosome-specific repeated DNA.