Any site characterisation relies on at least some laboratory tests, and some of those test results (often from basic tests such as Liquid Limit and Plastic Limit) are used in initial design calculations based on correlations. However, the reliability of those correlations is heavily dependent on the uncertainty in the results of those laboratory tests. There is an inter-laboratory proficiency testing scheme that has been running for over fifteen years in the UK, with many worldwide participants. This paper presents a compilation of the scheme’s findings to allow an assessment to be made regarding the reliability of different tests. As an example, repeatability within a single laboratory for the Liquid Limit test by cone penetrometer has been shown to be ± 1 %, but between different laboratories this has risen to ± 6 %. Similar ranges have been found in the Plastic Limits which, taken together with the uncertainties from the Liquid Limits, could give rise to significant concerns over using correlations based on the Plasticity Index. Other examples of results from various test methods are given in this paper and it is argued that much of the uncertainty comes not from the test method itself, but from other factors including basic equipment maintenance, calibration, technician training and competence. It will be seen that laboratory proficiency testing schemes are crucial in highlighting these problems and giving an opportunity to allow better assessment of the quality of both test results and, arguably, the laboratories that produce them.