A pilot site for CO2 storage in coal seams was set-up in the Upper Silesian Coal Basin in Poland in the scope of the RECOPOL project, funded by the European Commission. About 760 tons CO2 were injected into the reservoir from August 2004 to June 2005. Breakthrough of the injected CO2 was established, which resulted in the production of about 10% of the injected CO2 in this period. This paper reports on activities performed under the European Commission project MOVECBM that aimed at the assessment of the storage performance of the reservoir in the follow-up period, i.e. whether the injected CO2 was adsorbed onto the coal or whether it was still present as free gas in the pore space. The injection well was used for this purpose, as the production well had to be abandoned for permitting reasons. Several operational periods can be defined between the last injection in June 2005 and the abandonment of the well in October 2007. In the first period the well was shut-in to observe the pressure fall-off, from about 15.0 MPa at the wellhead after the last injection until about 4.5 MPa at the end of 2005. This pressure fall-off curve showed that the reservoir permeability was very low. This seemed to confirm the observed swelling of the coal during the injection period. In the first months of 2006 the pressure at the wellhead was decreased by releasing gas in a controlled way. The amount and composition of the gas were measured. As a result of the pressure reduction, the well flooded with water. A production pump was placed on the former injection well, enabling active production from the coal from March to September 2007. Results of these operations showed that whereas the gas production rates were as expected based on the experience with the production well, the water production was remarkably low. This could be related to permeability issues or, alternatively, indicate a drying effect of the CO2 in the reservoir. Further, the gas composition showed a predominance of CO2 over CH4 during the gas release that changed gradually into a predominance of CH4 over CO2 during the production phase. Although stabilization was not reached within the given production period, the composition approached a 60% methane, 40% CO2 ratio. This indicates that the exchange of these gases is more complex than often envisaged. After removal of the pump the well was filled with water, which ceased the gas release. This indicates that the pressure in the reservoir was back to its original, hydrostatic, state. As the total volume of CO2 produced was only a fraction of the amount that was injected, it can be concluded that the CO2 was taken up by the coal and is currently adsorbed. This gives confidence in the long-term stability of the injected CO2. © 2008 Elsevier B.V. All rights reserved.