Firms face a great deal of uncertainty in developing new technology. This uncertainty poses much more serious problems when development projects are new to firms and there are few historical data applicable to the development process in point. Examines operations-improvement processes in two shipbuilding firms and proposes an extended version of parallel R&D theory, called learn-ing-updating strategy. Based on a field study, presents empirical evidence for the effectiveness of the learning-updating strategy in allocating managerial resources for operations improvement.