<p>During the excavation process of deep hard rock tunnels, precutting rock with an abrasive water jet can weaken their strength and improve the efficiency of mining machinery. However, owing to the complex geological environment, abrasive jets cannot fully utilize their rock-cutting performance. To fully exploit the advantages of high-pressure abrasive water jets, five orthogonal experiments were designed for rocks with significant differences in strength. Experimental research has been conducted on the performance of rotating abrasive waterjet-cutting rocks. Moreover, a neural network prediction model for predicting rock-cutting characteristics is established by comprehensively considering rock mechanics parameters and abrasive water jet parameters. The results show that the cutting depth of rocks with different strengths increases nonlinearly with increasing work pressure of the abrasive water jet. The cutting depth decreases exponentially with increasing cutting velocity. The cutting depth first increases and then decreases with increasing target distance, and the best target distance is between 4 mm and 6 mm. The effect of the target distance on the cutting width of rock is the most significant, but the cutting width of high-strength rock is not sensitive to changes in the working parameters of the abrasive water jet. The average relative errors of BP (backpropagation) neural networks optimized by global optimization algorithms in predicting rock cutting depth and width are 13.3% and 5.4%, respectively. This research combines the working characteristics of mining machinery to study the performance of abrasive waterjet rotary cutting of rocks and constructs a predictive model for the performance of abrasive waterjet cutting of rocks that includes rock strength factors. This provides a new solution for quickly adjusting the working parameters of abrasive water jets according to mining conditions.</p>