Ozone observations from the Solar Backscatter UltraViolet/2 (SBUV/2) instruments and/or the Earth Probe Total Ozone Mapping Spectrometer (EP TOMS) have been assimilated in near-real time at NASA's Data Assimilation Office (DAO) since January 2000. The ozone data assimilation system was used as a tool for detecting and characterizing changes in the observation errors. The forecast model captures the geophysical variability. A change in the observed-minus-forecast (O-F) residuals, which are defined as differences between the incoming ozone observations and the collocated short-term model forecast, indicates a change in the assimilation system. If the model and the statistical analysis scheme are stable, then it points to a modification in instrument characteristics or a retrieval algorithm. However, sometimes a change in the ozone O-F residuals is caused by differences in the availability of the meteorological observations or modifications in the meteorological assimilation system whose winds are used to drive the ozone transport model. The O-F residuals are routinely produced and monitored in the assimilation process. Using examples from the NOAA 14 and NOAA 16 SBUV/2 instruments, and the EP TOMS, we demonstrate that the monitoring of time series of O-F residual statistics is an effective, sensitive, and robust method for identifying time-dependent changes in the observation-error characteristics of ozone. In addition, the data assimilation system was used to assist in the validation of updated calibration coefficients for the NOAA 14 SBUV/2 instrument. This assimilation-based monitoring work is being extended to ozone data from instruments on new satellites: Environmental satellite (Envisat), Earth Observing System (EOS) Aqua, and EOS Aura.