Deep, high-stress soft-rock tunnels are prone to sustained deformation during long-term operation; even following repairs to conventional support systems, rock mass failure and support component failure may still occur. In response to issues such as the persistent occurrence of deformation and failure in deep tunnels following several rounds of repairs to the original support scheme, an analysis of rock mass deformation and failure was conducted. A new blasting-induced stress relief repair scheme was introduced and field trials were conducted. On this basis, the anchored rock mass was modelled as a load-bearing arch. Taking into account the tunnel’s geometric parameters, rock mass strength, anchor support parameters and the propagation and attenuation characteristics of blasting stress waves, a mechanical model of the anchored load-bearing structure under dynamic loads from footwall blasting was established, and a mechanical criterion for the stability of the anchored load-bearing capacity under blasting-induced stress relief was derived. The research findings indicate that: 1) Following the refurbishment of the original scheme, deformation and failure continued to occur; the zone of severe damage was located an average of 3.44 m from the tunnel, primarily because the original support scheme had failed to address bottlenecks such as the sustained application of high stresses, thereby preventing the tunnel from maintaining long-term stability; 2) Following the implementation of the blasting pressure relief repair scheme, a fracture zone with a radius of approximately 0.6–1 m was formed. Vibration monitoring data remained within the specified limits, and no vibration-induced impact damage occurred in the tunnel. The reduction rates in displacement for the side walls, shoulders, floor and roof were 75.85 per cent, 60.92 per cent, 59.76 per cent and 81.15 per cent respectively, indicating that the surrounding rock was effectively stabilised; 3) Analysis of case studies on influencing factors indicates that the stability mechanical criterion η is positively correlated with the cohesion of the surrounding rock, anchor length, working load of the anchors and blasting depth, and negatively correlated with the equivalent radius of the tunnel, anchor spacing and peak blasting vibration. Among these factors, anchor length has the most significant influence; 4) The range of variation in the mechanical criterion η differs for different roadways. Using the basic parameters in this paper as a reference standard, the main range of variation for the mechanical criterion η is (0, 0.2), with a benchmark reference value of approximately 0.15. This range can be divided into four intervals: the severely affected zone (0, 0.05), the moderately affected zone (0.05, 0.1), moderate impact zone (0.1, 0.15) and mechanically stable zone (0.15, 0.2), and corresponding measures have been proposed. The research findings can provide guidance for subsequent blasting stress relief design.