Air parcel photochemical trajectory and 0-D box photochemical models have proven to be invaluable for analyzing data collected from aircraft platforms by the GTE (Global Tropospheric Experiment). The major shortcoming’s of this approach have been our relative inability to prescribe the initial conditions with confidence and possible errors in air parcel trajectory histories and related physical and chemical parameters. We are currently developing a tangent linear model (TLM)/adjoint trajectory-photochemical model that better reflects these uncertainties and is also suitable for assimilating chemical data corresponding to the air parcel history. Such data could be available from satellite platforms in the near future. We have used ADIFOR, a software programming tool developed at Argonne National Laboratory and Rice University, to develop the TLM. Initial results indicate that the ozone mixing ratios in the TLM could diverge substantially from those in the basic non-linear model (NLM) within a day of initialization, for moderate mixing ratios of NOx in the plume. However, the mixing ratios for CO calculated with TLM are in better agreement with the basic NLM results for the different NOx conditions simulated in the plume. This observation indicates that additional difficulties arising from the strongly nonlinear character of trace gas chemistry, particularly in urban plumes, need to be considered in devising data assimilation strategies for air pollution models.