In this paper the population for male of rural area in Bangladesh is predicted by using the geometric growth rate method. The predictions are computed in a three-step procedure. In the first step, the prediction are computed using an exponentail model estimated by Quasi-Newton method for the years 1974, 1981, 1991, and 2001 using the package STATISTICA. Using the cross-validation predictive power (CVPP) criterion and coefficient of determination, the shrinkage coefficient $lambada$  is constructed. The shrinkage coefficient determines the adequacy of the first step prediction. In the second step, these predicted values are used to estimate the growth rate for different age groups by using the geometric growth arte method. In the final step, considering the population for male of rural area in Bangladesh for the census year 2001 as the base population and using the estimated geometric growth rate of the second step estimation, the predictions for the male population of rural area of Bangladesh are estimated for the years 2002 through 2031 by applying geometric growth rate method.