An inter-laboratory study of high-pressure gas sorption measurements on two carbonaceous shales has been conducted in order to assess the reproducibility of the sorption isotherms and identify possible sources of error. The measurements were carried out by seven international research laboratories using either in-house or commercial sorption equipment (manometric and gravimetric methods). Excess sorption isotherms for methane, carbon dioxide and ethane were measured at 65 C and at pressures up to 25 MPa on two organic-rich shales in the dry state. The samples used in this study were taken from immature Posidonia shale (Germany) and over-mature Upper Chokier Formation (Belgium). Their total organic carbon (TOC) contents were 15.1% and 4.4%, respectively, and their vitrinite reflectance (VRr) values 0.5% and 2.0%. The objective of this study was to assess the reproducibility of sorption isotherms among laboratories each following their own measurement and data reduction procedures. All labs were asked to follow a predefined sample drying procedure prior to measurement in order to minimize any effects related to moisture. The reproducibility of the methane excess sorption isotherms was better for the high-maturity shale (within 0.02-0.03 mmol/g) than for the low-maturity sample (up to 0.1 mmol/g), similar to observations in earlier inter-laboratory studies on coals. The reproducibility for CO2 and C2H6 sorption isotherms was satisfactory at pressures below 5 MPa, however, the results deviate considerably at higher pressures. Anomalies in the shape of the excess sorption isotherms were observed for CO2 and C2H6 and these are explained as being due to high sensitivity of gas density to temperature and pressure close to the critical point as well as from a limited measurement accuracy and possibly uncertainty in the equation of state (EoS). The low sorption capacity of carbonaceous shales (as compared to coals and activated carbons) sets very high demands on the accuracy of pressure and temperature measurement and precise temperature control. Furthermore, the sample treatment, measurement and data reduction procedures must be optimized in order to achieve satisfactory inter-laboratory consistency and accuracy. Systematic errors must be minimized first by calibrating the pressure and temperature sensors to high-quality standards. Blank sorption measurements with a non-sorbing sample (e.g. stainless steel) can be used to identify and quantitatively account for measuring artifacts resulting from unknown residual systematic errors or from the limited accuracy of the EoS. The possible sources of error causing the observed discrepancies are discussed. (C) 2014 Elsevier B.V. All rights reserved.