While the technologies for autonomous on-orbit servicing of satellites are emerging, no general conclusions have yet been drawn regarding the cost-effectiveness of on-orbit servicing for space systems. The first part of the paper considers a general model that applies the traditional approach to onorbit servicing to a large trade space of space missions and servicing infrastructures. Two major limitations of this approach are demonstrated. First, the servicing cost uncertainty is shown to be too high to yield meaningful conclusions. Second, this approach does not take into account the flexibility provided by on-orbit servicing to space systems. The second part of the paper proposes a new perspective on on-orbit servicing. The value of servicing, defined as the maximum price under which on-orbit servicing would be interesting, is studied independently from its cost. A framework to quantify the value of the flexibility provided by on-orbit servicing to space systems is developed. The application of this framework should prove very useful in identifying the space missions for which on-orbit servicing would offer the most potentials. The results should serve as a guide to future technology development.