Continuous self-baking electrodes, i.e. Soderberg electrodes, are commonly employed in the industrial smelting of ores in submerged arc furnaces. These electrodes are formed from electrode paste that is baked during operation into solid carbonaceous and ultimately graphite electrodes. Coal tar pitch (CTP) is used as a binder in the electrode paste. In order to ensure the production of quality electrode paste, CTP is analysed for softening point (SP), coking value (CV), as well as toluene- (TI) and quinoline insoluble (QI) contents, together with proximate analysis for the determination of fixed carbon, volatile, ash and moisture contents. Ultimate analyses for the determination of carbon, hydrogen, nitrogen, sulphur and oxygen contents are also performed. However, the TI and QI content analyses are time consuming and harmful solvents are used. In this paper it was proven that multi-linear regression (MLR) analysis can be used to derive equations to calculate TI and QI from other parameters, i.e. SP, CV, as well as proximate and ultimate analyses. Root mean square error differences between MLR calculated TI and QI contents for 12 CTP samples sourced from suppliers throughout the world were found to be approximately 0.36 and 032, respectively. Although it is not foreseen that calculation of these two parameters will replace experimental determinations, the MLR equations determined in this study can be used by CTP producers and consumers as proxies for TI and QI. This will reduce the use of harmful solvents and the frequency required to conduct time consuming experimental determinations for TI and QI. (C) 2015 Elsevier B.V. All rights reserved.