A method for assimilating observations of long-lived species such as ozone (O₃) and nitrous oxide (N₂O) in a three-dimensional chemistry transport model (3D-CTM) is described. The model is forced by the temperature and wind analyses from the European Centre for Medium-Range Weather Forecasts (ECMWF). The O₃ and N₂O fields used in this study are obtained from the Sub-Millimeter Radiometer (SMR) aboard the Odin satellite. The assimilation technique used is the sequential statistical interpolation approach. The parametrization of the error covariance matrix of the model forecast field is described. A sensitivity study of the system parameters is done in terms of the OMF (observation minus forecast) vector also called “innovation” vector and in terms of the χ² (chi-square) test. The effect of the correlation distances is critical for the assimilated field. The RMS (root mean square) of the OMF for the correlation distances is minimal for values of 1500 km in the meridional direction and 500 km in the zonal direction for both O₃ and N₂O. The treatment of the meridional distance as a function of latitude does not reveal an important improvement. The χ² diagnostic shows that the asymptotic value of the model error (the model error of saturation) is optimal for the value of 12.5% for O₃ and 18% for N₂O. We demonstrate the applicability of the developed assimilation method for the Odin/SMR data. We also present first results of the assimilation of Odin/SMR ozone and nitrous oxide for the period from 22 December 2001 to 17 January 2002.< Réduire