This paper examines the effect of correlation level on the use of auxiliary variable in double sampling for regression estimation. The findings revealed that double sampling for regression with high correlation coefficient (between the auxiliary and study variables) has the minimum variance