A Fast Delivery (FD) system has been developed to provide GOME (Global Ozone Monitoring Experiment) total ozone columns within 3 h after observation. This meets the growing demand for near real-time ozone products for applications such as the assimilation of ozone in Numerical Weather Prediction (NWP) models, monitoring the status of the ozone layer and improving surface UV radiation forecasts. The FD system uses only a small part of the raw GOME data that are available in near real-time. The FD level 0 to 1 and level 1 to 2 processors are based on algorithms as used within the off-line ESA/DLR GOME Data Processor (GDP). However, several changes in the algorithms have been made to improve the retrieval of the ozone columns: (1) the use of effective ozone absorption cross-sections based on ECMWF (the European Centre for Medium-Range Weather Forecasts) temperature forecast in the ozone slant column fit, (2) the use of the DAK radiative transfer model and the Fortuin and Kelder ozone climatology for the air mass factor calculations and (3) the use of the FRESCO (Fast Retrieval Scheme for Cloud Observables) cloud algorithm. The accuracy of the FD ozone columns has been determined by a pseudo-global validation study involving ground-based total ozone observations from about 15 NDSC (Network for the Detection of Stratospheric Change) sites. There is a reasonable general agreement with the ground-based measurements (within 5% for low and mid latitudes, within - 10 to + 5% for high latitudes) and the level of quality meets the requirements for near real-time applications.