The empirical stope design techniques developed by Mathews et al (1981) and refined by Potvin (1988) and others, have gained rapid acceptance in Australia in the 1990s as a simple, first-pass' means of designing primary stopes. However, it is recognised that these stability graph' methods can only provide broad design guidelines. Recently published stability graphs (Hutchinson and Diederichs, 1996) encompass a wide range of geological environments, which are not discussed, and the influence of time and blasting practices cannot be easily reviewed. The scatter' in the graphs must reflect, to some degree, different operating practices as well as subtle but important differences in the pre-mining conditions that are not adequately accounted for in the modified stability number'. Development of a local', site-specific stability graph may overcome some of these problems. At Normandy's Golden Grove Operations, Western Australia, detailed reviews of stope stability performance, including Cavity Monitoring System (CMS) surveys, are undertaken for every stope, with the objective of developing a series of local stability graphs, customised to site conditions and calibrated in terms of Average Falloff Thickness' (AFT). Cablebolting of stope hangingwalls is used to develop reinforced beams' on each sublevel, effectively dividing the full span into smaller, more stable subspans'. By analysing stope stability performance in terms of fall-off, cost-benefit studies can be undertaken of the effect of cablebolting and other options for modifying stope spans. The approach used to develop the local stability graphs, and their integration into routine stope design procedures at Golden Grove, is described. The experience gained during stope extraction in the Scuddles orebodies has been successfully applied to the Gossan Hill C-Panel' orebody. This methodology should be applicable to other underground mining operations.