This paper describes a novel feed-hopper level estimation and control scheme for addressing the known problem of unreliable and occasionally corrupted feed-hopper level measurement in a cone crusher. The approach involves estimating the feed-hopper level with an adaptive time-variant state estimator. The proposed adaptive scheme delivers asymptotically unbiased feed-hopper level estimates, despite using an inherently biased state estimator with biased measurement(s) and/or model, and therefore addresses the common pitfall of state estimators. The paper details the entire control system design procedure, from the fundamental theory, through dynamic modeling and estimator/controller tuning, to the design validation and control performance evaluation. The performance of the proposed scheme is evaluated through extensive full-scale tests in various production scenarios, including process start-up, level setpoint changes, and mass flow disturbance rejection. The full-scale tests revealed a number of benefits compared to the straightforward level control implementation. These benefits include the possibility of recovering from a temporary loss of measurement signal, smaller control effort, and increased system. robustness due to an increased ability to withstand measurement errors. Therefore, the proposed scheme will enable more consistent size reduction and provide protection against performance degradation and process down-time.