A model to assimilate stratospheric chemical constituents is proposed. The chemical transport model used is two dimensional on an isentropic surface and solves the photochemical equations using an extended family approach. Thirty-day integrations are performed at the three isentropic surfaces given by the potential temperature values 600 K, 850 K and 1400 K starting on January 1, 1979. During the integrations, Limb Infrared Monitor of the Stratosphere (LIMS) satellite data are assimilated into the model and the results are compared with independently mapped (map archive tape (MAT)) data. The assimilation results are found to be in good agreement with the independent analyses at all three levels. Preliminary assimilations revealed significant biases in modeled HNO3 at 850 K and in O3 at 1400 K. However, by changing the adjustable parameters of the assimilation model the HNO3 bias was successfully eliminated. The O3 bias was easily eliminated by the addition of an extra source term. The comparisons reveal possible minor weaknesses in the MAT analyses. First, the chemical constituents appear to be occasionally in error in the MAT data during rapidly changing dynamical events such as stratospheric warmings. Second, the temporal continuity is sometimes lacking in the MAT analyses, leading to rapidly changing concentrations which cannot be understood on the basis of conventionally understood photochemistry. On the other hand, the assimilation model used here does not produce very satisfactory analyses when the chemical data are very noisy. The differences between the assimilated and MAT analyses are quite small and are generally less than data errors. Although the assimilation model is here applied to LIMS data, the method is generally applicable to any data source and could be especially useful for analyzing the large volume of data from the Upper Atmosphere Research Satellite. ¿ American Geophysical Union 1992