Nowadays, the health prognosis is popularly recognized as a significant lever to improve themaintenance performance of modern industrial systems. Nevertheless, how to efficiently exploit prognostic informationfor maintenance decision-making support is still a very open and challenging question. In this paper,we attempt at contributing to the answer by developing a new parametric predictivemaintenance decision frameworkconsidering improving health prognosis accuracy. The study is based on a single-unit deteriorating systemsubject to a stochastic degradation process, and to maintenance actions such as inspection and replacement.Within the new framework, the system health prognosis accuracy is used as a condition index to decide whetheror not carrying out an intervention on the system. The associated mathematical cost model is also developedand optimized on the basis of the semi-regenerative theory, and is compared to a more classical benchmarkframework. Numerical experiments emphasize the performance of the proposed framework, and confirm theinterest of introducing the system health prognosis accuracy in maintenance decision-making.