Modern underground mining methods often call for the excavation of lar
ge, often unsupported voids called stopes. In many cases, these are no
nentry stopes, with the removal of broken rock (mucking) performed by
remote control. With the risk to human safety thus reduced, stability
concerns within the stopes are primarily driven by the economics of un
planned overbreak and waste rock dilution of the mined ore. In this co
ntext, it is appropriate to speak of acceptable risks of instability a
nd to optimize design by balancing risk costs with productivity potent
ial. Larger individual stopes generally increase productivity while re
ducing stability. The modified stability graph is a popular empirical
mine design tool that has been calibrated to provide discrete recommen
dations for maximum stope dimensions based on a rock quality index. It
seems appropriate to extend the method to account for variability in
the calibration data and also for variability in each new set of input
parameters. A risk template and associated probabilistic techniques a
re proposed to supplement the discrete design limits currently establi
shed.