The CHaracterizing ExOPlanet Satellite (CHEOPS), to be launched in December 2019, will detect and characterize small size exoplanets via ultra high precision photometry during transits. CHEOPS is designed as a follow-up telescope and therefore it will monitor a single target at a time. The scientific users will retrieve science-ready light curves of the target, automatically generated by the CHEOPS data reduction pipeline of the Science Operations Centre. This paper describes how the pipeline processes the series of raw images and, in particular, how it handles the specificities of CHEOPS data, such as the rotating field of view, the extended irregular Point Spread Function, and the data temporal gaps in the context of the strict photometric requirements of the mission. The current status and performance of the main processing stages of the pipeline, that is the calibration, correction and photometry, are presented to allow the users to understand how the science-ready data have been derived. Finally, the general performance of the pipeline is illustrated via the processing of representative scientific cases generated by the mission simulator.