We discuss in this work the first-order, second-order and pseudo-second-order estimations of Lagrange multipliers in nonlinear constrained minimization. The paper also justifies estimations and strategies that are used by two nonlinear programming algorithms that are also briefly described.