When using experimental data and/or In Silico predictions, it is important to assess the quality of the information as part of an overall assessment of confidence. When assessing the reliability of experimental data, it is usual to consider the reproducibility of the test performed by comparing the study design and execution against published and accepted test guidelines and the extent to which this information is documented. The Klimisch score is a simple and widely used method to quantify the reliability of experimental data based on a score of 1-4, as summarized below1:

1Reliable without restriction• Well documented and accepted study or data from the literature • Performed according to valid and/or accepted test guidelines (e.g., OECD) • Preferably performed according to good laboratory practices (GLP)
2Reliable with restriction  • Well documented and sufficient  • Primarily not performed according to GLP  • Partially complies with test guideline
3Not reliable• Interferences between the measuring system and test substance • Test system not relevant to exposure • Method not acceptable for the endpoint • Not sufficiently documented for an expert review
4Not assignable• Lack of experimental details • Referenced from short abstract or secondary literature

In Silico results are being increasing used for hazard and risk assessment alongside experimental data and hence it is important to assess the reliability of both types of results within a harmonized grading scheme. Such a score, called the Reliability Score, was generated as part of the in silico toxicology protocols project.2 This score was created by a team of over 80 experts in the field and reflects the current state of the art. It considers that there is an increase in reliability when multiple and complementary models are used that predict the same results. In addition, a further increase in reliability can be achieved by performing an expert review of the totality of information, including any output from the in silico models (e.g., results for chemicals analogs, important QSAR descriptors, proposed mechanism(s) of action). Where such a review increases the reliability, the Reliability Score is adjusted accordingly. A read-across assessment, which includes a similar type of expert review, is considered in this same category. The Reliability Score is a five-point grading that extends the Klimisch score to include In Silico model results as well the aforementioned factors that increase the reliability. The following table summarizes the Reliability Score alongside the Klimisch score:

Reliability ScoreKlimish ScoreDescription
RS11Data reliable without restriction
RS22Data with restriction
RS3Expert review of in silico result (s) and/or Klimisch 3 or 4 data
RS4Multiple concurring prediction results
RS5Single acceptable in silico result
RS53Data not reliable
RS54Data not assignable

Reliability is an important component of any confidence assessment for a toxicological hazard outcome, alongside an evaluation of the relevance and completeness of the information. A framework for assessing such a confidence measure is described in a number of recent publications.2,3

Please do not hesitate to get in touch if you wish to discuss such approaches with a scientist at Instem (info@instem.com).


  1. Klimisch, H.-J., Andreae, M., Tillmann, U., 1997. A systematic approach for evaluating the quality of experimental toxicological and ecotoxicological data. Regul. Toxicol. Pharmacol. 25, 1–5. http://dx.doi.org/10.1006/rtph.1996.1076
  2. Myatt et al., In silico toxicology protocols, Regul. Toxicol. Pharmacol. 96 (2018) 1–17. https://doi.org/10.1016/j.yrtph.2018.04.014
  3. Johnson et al., Evaluating confidence in toxicity assessments based on experimental data and in silico predictions, Comput. Toxicol. 21 (2022) 100204. https://doi.org/10.1016/j.comtox.2021.100204

Published by Glenn Myatt

Glenn J. Myatt is the co-founder of Leadscope and currently Senior Vice President, In Silico & Translational Science Solutions at Instem with over 30 years’ experience in computational chemistry/toxicology. He holds a Bachelor of Science degree in Computing, a Master of Science degree in Artificial Intelligence and a Ph.D. in Chemoinformatics. He has published 37 papers, 11 book chapters and three books.