<p>To reduce the uncertainty associated with the traditional definition of tunnel boring machine (TBM) utilization (<i>U</i>) and achieve an effective indicator of TBM performance, a new performance indicator called rock mass-related utilization (<i>U</i>_r) is introduced; this variable considers only rock mass-related factors rather than all potential factors. This work aims to predict <i>U</i>_r by adopting the rock mass rating (RMR) and the moisture-dependent Cerchar abrasivity index (CAI). Substantial <i>U</i>_r, RMR and CAI data are acquired from a 31.57 km northwestern Chinese water conveyance tunnel via tunnelling field recordings, geological investigations and Cerchar abrasivity tests. The moisture dependence of the CAI is explored across four lithologies: quartz schists, granites, sandstones and metamorphic andesites. The potential influences of RMR and CAI on <i>U</i>_r are then investigated. As the RMR increases, <i>U</i>_r initially increases and then peaks at an RMR of 56 before declining. <i>U</i>_r appears to decline with CAI. An investigation-based relation among <i>U</i>_r, RMR and moisture-dependent CAI is developed for estimating <i>U</i>_r. The developed relation can accurately predict <i>U</i>_r using RMR and moisture-dependent CAI in the majority of the tunnelling cases examined. This work proposes a stable indicator of TBM performance and provided a fairly accurate prediction method for this indicator.</p>