Many formulae have been suggested to estimate the ridge parameters. Most of these formulae ignore the features of the model such as number of the explanatory variables, degree of illconditioning and the significancy of the model. In this article two formulae are proposed for selecting the ridge parameter K; for ridge regression. A Monte Carlo study is conducted to compare the mean square errors of ridge regression under the new formulae and some other formulae. The numerical results of the simulation indicate that the performance of the new formulas does produce the smaller mean square errors.