A panel cave using LHDs for ore extraction came into production at Premier in 1990 at a depth of 630 metres below surface on the western side of the mine in the BA5 mining block. A post-undercut mining sequence was used with all development on both the undercut and production levels completed prior to the undercut being advanced over the pre-developed drawbells. Support and rock mass damage on the production level was extensive as the undercut moved overhead. Undercutting and coarse fragmentation created problems. The mining sequence was changed to an advance undercut with the minimum of development and support undertaken on the production level prior to the undercut moving overhead. Only once the abutment stresses had stabilised were drawbells developed and final support installed. The height of the undercut drill design was decreased from 18 to 13 metres. The changes resulted in markedly less rock mass and support damage on the production level. Lessons learned in the BA5 mining block on the 630 level were incorporated into planning and scheduling the BB1E mining block which came into production on the 732 metre level on the eastern side of the mine in 1996. An advance undercut mining sequence was planned, undercut height was minimised and undercutting parameters were optimised in terms of the BA5 experience. Numerous problems still occurred. As the undercut area increased in size severe spalling from the cave back resulted in vertical propagation of the cave to a height of 30 metres instead of the planned 13 metres. The resultant coarse fragmentation created secondary blasting problems and an excess of undercut ore that had to be loaded and trammed, slowed the rate of undercut advance. The planned mining sequence was compromised and considerable support and rock mass damage resulted on the production level below. Once these problems had been overcome the undercut advanced at the planned rate and continuous caving initiated at the predicted hydraulic radius. At a later stage the unexpected ingress of water hampered undercut drilling and slowed the rate of undercut advance in a section of the cave. Leads and lags between adjacent tunnels and the overall face shape could not be developed as planned. The rate of natural horizontal cave propagation exceeded the rate of undercut advance and three undercut drilling tunnels were extensively damaged. More tunnel development on the production level prior to undercutting than was planned resulted. The slow advance of the undercut and high abutment stresses caused extensive damage to installed support and the rock mass on the production level, 15 metres below. Experience in the BB1E mining block on the 732 metre level has taught that an advance undercut mining sequence reduces damage on the production level markedly. Extremely tight control on the rate of undercut advance, leads and lags between adjacent tunnels and overall undercut face shape as well as extraction ratio, tunnel development and support sequence on the production level below is required. Premier Mine plans to implement a block cave at a depth in excess of 1000 metres within the next decade. All the experience gained on current mining levels together with data collection and numerical modeling is being used to plan the proposed mining operation. The geotechnical challenge of mining at this depth in weak rock has been identified as the greatest risk to the project. Challenges centre around adopting the correct mining sequence, limiting the extent of rock mass damage associated with undercutting and effective support. The hydraulic radius needed to initiate caving, fragmentation, drawpoint spacing, secondary blasting and the possibility of seismicity are geotechnical concerns that are being addressed. Results to-date suggest that a pre-undercut mining sequence with all development and support on the production level carried out only after the undercut abutment stresses have passed overhead will have to be implemented.