<p>The rapid evolution of advanced manufacturing has heightened interest in machinery prognostics under time-varying operating conditions (TVOCs), which pose significant challenges to safe and reliable operation. This review summarizes recent advancements in machinery prognostics under TVOCs. First, we discuss experimental studies of machinery degradation using data from bearings, components of machine tools, and aero-turbine engines. Second, health indicators are constructed for anomaly detection and degradation assessment, detailing methodologies based on statistical parameters, signal processing, and machine learning to capture underlying degradation patterns. Third, we explore remaining useful life prediction methods, including direct (deep learning and regression models) and indirect (degradation modeling and similarity analysis) mapping approaches. Theoretical foundations and algorithmic flows are outlined. This review could serve as a comprehensive guide, highlighting state-of-the-art techniques and future directions in this advancing field.</p>