From the Hubble Space Telescope (HST) example, a model of a serviceable scientific mission is developed to study on-orbit repair and upgrading, including spacecraft repair, payload instrument upgrading, and bus subsystems upgrading. Mission utility is measured by the onboard payload instrument discovery efficiency. A Monte Carlo simulation is used to model uncertainty in the arrival of new technologies, random spacecraft failures, and catastrophic failures of a servicing operation. The value of the option to repair. and upgrade, as well as the impact of different design choices, is investigated using actual data from the HST manned servicing missions. The main differences between manned and unmanned servicing missions are then analyzed in an attempt to derive conclusions for robotic on-orbit servicing of scientific missions. In light of the cases examined, designing for serviceability exclusively for satellite repair does not seem very valuable. In contrast, upgrading can significantly increase mission utility especially if the technology embedded in the serviceable modules is evolving rapidly. Technology upgrades via on-orbit servicing have been identified as a very promising concept. Large increases in mission utility can be realized for a cost significantly lower than the cost of replacing the whole satellite.