Appropriate rock characterization is beneficial in providing a reliable judgment on rock properties which is crucial for the design process of rock engineering applications. However, it can be difficult to obtain consistent mechanical parameters due to substantial variations in rock properties. In this research, uniaxial compression tests on dolerite specimens collected from a gold mine in Western Australia showed substantial scatter in the results. Rock categorization based on the P-wave velocities is as ac-curate as the thin section analysis, which suggests that they can be used together to gain a more accurate initial understanding of the rock types before any laboratory testing. The quality of specimen preparation and rock-machine interaction greatly affect the test results. For instance, non-parallelness of loading platens can lead to considerable scatter of the testing results, which would be perceived as rock vari-ability. It is suggested that the current testing standards should be modified towards a better control of the loading machine performance and equipment precision. Finally, the possibility of pre-existing microcracks in rock, neither detected by the thin section analysis nor by the ultrasonic measurement, must be examined by computed tomography (CT) scanning as they can affect the test results. This study will enhance our knowledge about the sources of variability in laboratory test results of rock which is essential for obtaining reliable data.