Uniaxial Compressive Strength (UCS) and sonic velocity correlations are used widely in the Australian coal mining industry to predict in situ rock strength. These models are cheap, fast and easy to produce, as well as easy to understand and have a number of practical applications in mine planning and design. The major downfall of these models is that there is a large variation in UCS values at high sonic velocities limiting their predictive ability. The aim of this research project is to improve the reliability of UCS/Sonic velocity correlations by reducing the variability in the underlying data. This is performed by identifying and eliminating sources of error affecting the data and looking at the impact of certain factors on the quality of the correlations. Results show that improved models can be obtained by filtering the datasets to remove samples with high length-to-height ratios, conglomerate or pebbly lithologies, and large sonic velocity ranges.