The measurement of soil moisture can be a time-consuming task that can be difficult to capture spatially and temporally. The accuracy of soil moisture measurements is essential to improve aspects of hydrology in a range of modelling situations. This paper compares soil moisture measurements from two un-calibrated in-situ measurement devices against gravimetric data. The devices used are a Delta T Theta Probe and a Campbell Scientific CS659 while the gravimetric readings are from soil cores (12 cm and 21 cm). The soil moisture readings were taken over two large semi-arid catchments (562 km² and 808 km²) located in South East Australia in the Hunter Valley, NSW. Multiple field campaigns were conducted in 2014, 2015 and 2018, resulting in 308 gravimetric samples for analysis in predominately clay soils. The two core depths sampled showed a strong correlation coefficient (R value) of 0.89. The gravimetric and probe measurements returned R values of approximately 0.8 for 2014 and 2015. The 2018 results showed a decrease in correlation (to approximately 0.3 and 0.5) although this coincided with average gravimetric soil moisture values being much lower than previous data collection campaigns (approximately 13% opposed to 20-23%). The extreme dry period potentially the reason for the reduced correlation. The manufacturer calibrated probe measurements did not provide a 1:1 relationship with the lab based gravimetric soil moisture. Results show that either the Theta Probe or CS659 are comparable to the gravimetric results in most conditions. Both the Theta Probe and CS659 regressions produced root mean square errors that were within the quoted accuracy in the device manuals, ±5% and ±3% respectively. The instruments may be used in conditions showing soil moistures of ~5% to ~45%, although the best results will be obtained by using appropriate techniques and knowing the potential limitations of devices. The linear regression equations found in this study may also allow calibration of probes for future measurements.