This paper deals with the problem of identifying the inertia parameters of a manipulator. We begin by introducing the terminology of minimal linear combinations of the inertia parameters (MLCs) that are linearly independent of one another and determine the manipulator dynamics while keeping the number of linear combinations of the inertia parameters to a minimum. The problem is then to find an identification procedure for estimating the MLCs and to use the MLCs in the inverse dynamics for control. The recursive Newton-Euler formulation is rederived in terms of the MLCs. The resulting formulation is almost as efficient as the most efficient formulation in the literature. This formulation also provides a starting point from which to derive a recursive identification procedure. The identification procedure is simple and efficient, since it does not require symbolic closed-form equations and it has a recursive structure. The three themes concerning the dynamic modeling of a manipulatorthe MLCs, the inverse dynamics in terms of the MLCs, and the identification procedureare treated in sequence in this paper.