Abstract. We describe a sequential assimilation approach useful for assimilating tracer measurements into a three-dimensional chemical transport model (CTM) of the stratosphere. The numerical code, developed largely according to Kha00, uses parameterizations and simplifications allowing assimilation of sparse observations and the simultaneous evaluation of analysis errors, with reasonable computational requirements. Assimilation parameters are set by using χ² and OmF (Observation minus Forecast) statistics. The CTM used here is a high resolution three-dimensional model. It includes a detailed chemical package and is driven by UKMO (United Kingdom Meteorological Office) analyses. We illustrate the method using assimilation of Upper Atmosphere Research Satellite/Microwave Limb Sounder (UARS/MLS) ozone observations for three weeks during the 1996 antarctic spring. The comparison of results from the simulations with TOMS (Total Ozone Mapping Spectrometer) measurements shows improved total ozone fields due to assimilation of MLS observations. Moreover, the assimilation gives indications on a possible model weakness in reproducing polar ozone values during springtime.