A two-stage study was undertaken to establish what effect, if any, perchloroethylene has on coal coking properties. Stage 1 was a kinetic study in which the Gieseler fluidity of samples of a high fluidity coal was measured over a 64-week period. Two of the samples were left untreated, with one stored in air and one immersed in water. The third sample was also stored in air, but was soaked in perchloroethylene for 2 h and then air dried immediately prior to fluidity testing. In Stage 2, seven different coking coals were studied, covering three different Australian coal measures and a wide range of coking rank. For each coal, a 40 kg clean coal composite was formed using laboratory water-based methods, a Reflux Classier for the 0.038-0.25 mm and 0.25-2.0 mm size fractions and a Mintek jig for the 2.0-16 and 16-50 mm size fractions. This 40 kg composite was then subdivided into four sub-samples. Samples A and B were coked immediately, whereas Samples C and D were stored in open trays at ambient conditions for a month before coking. Samples B and D were soaked in perchloroethylene for 2 h and then air dried prior to coke testing, whereas Samples A and C were left untreated. Coal properties were measured prior to coking, and after coking the NSC Coke Reactivity Index ( CRI) and Coke Strength after Reaction ( CSR) were measured. In Stage 1 it was found that storage under water was able to maintain the sample fluidity better than refrigerated storage. The combined Stage 1 and 2 results showed that perchloroethylene had a detrimental effect on the coking properties of many coals. Hence coking tests on clean coal composites formed using heavy organic liquids may often under predict a coal's true coking properties. This could lead to many coal deposits being undervalued. Perchloroethylene had the largest detrimental effect on the fluidity, CRI and CSR of coals that started with relatively poor properties. Perchloroethylene had a negligible effect on coals with relatively good initial coking properties. There was no clear correlation with vitrinite reflectance. The three most seriously affected coals all had relatively high inertinite content. It is speculated that the higher porosity of inertinite might allow greater access and retention of perchloroethylene in the coal particles. The laboratory water-based methods were able to produce clean coal composites that matched the proximate, petrographic and coal swelling number properties of the coal preparation plant product. The dilatometer and fluidity results were lower than those of the coal preparation plant, but this can be explained by the up to three month delay between mining of the original sample and the coking of the composite. These delays were an unavoidable part of the research nature of this project and had nothing to do with the water based methods deployed. If these water-based methods of preparing clean-coal composites were to become standardized, then there is no reason why a well equipped coal testing facility could not receive a refrigerated bore core and produce a clean coal composite within the period of a week. (C) 2011 Elsevier Ltd. All rights reserved.