The ridge regression, as a modification of least squares, is one of the ways of the ways of overcoming multicollinearity of regressors in regression analysis. The central problem in the application of the ridge regression is the choice of the perturbation factor k. The paper compares the performance of some subjective (SM) and objective (OM) methods of selection k in respect of the estimates of the mean squared estimation error (MSEE) and the mean squared prediction error (MSPE). The chosen methods were applied on empirical data which relate to social product and some other relevant factors of agriculture of Yugoslavia and its regions.