Determining coal pillar strength equations from databases of stable and/or failed case-histories is more than 50 years old and has been applied in different countries by different researchers in a range of mining situations. Whilst common wisdom sensibly limits the use of the resultant pillar strength equations and methods to design scenarios that are consistent with the founding database, there are a number of examples whereby failures have occurred as a direct result of applying empirical design methods to coal pillar design problems that are inconsistent with the founding database. The paper explores the reasons as to why empirically-derived coal pillar strength equations tend to be problem-specific, and so should perhaps be considered as providing no more than a pillar strength “index”. These include the non-consideration of overburden horizontal stress within the mine stability problem, an inadequate definition of super-critical overburden behaviour as it applies to standing coal pillars and the non-consideration of overburden displacement and coal pillar strain limits, all of which combine to potentially complicate and so confuse the back-analysis of coal pillar strength from failed cases. A modified coal pillar design representation and model is presented based on coal pillars acting to reinforce a horizontally-stressed overburden, rather than suspend an otherwise unstable self-loaded overburden or section thereof, the latter having been at the core of historical empirical studies into coal pillar strength and stability.