The circulation of the north Atlantic as deduced from an inverse calculation is tested against the historical record of tritium. The circulation model is first used to do a ''forward'' calculation of the tritium transient using published estimates of atmospheric injection rates, and of plausible estimates of the tracer history at the open boundaries of the model. Major discrepancies are found between this calculation and the observations of interior distributions of tritium. An attempt is then made to remove these discrepancies by modifying the atmospheric injection rates and the initial estimates of open boundary time histories, treating these boundary conditions as control variables. The procedure requires answering difficult questions as to the accuracy with which atmospheric transfer to the ocean and the boundary conditions at the northern limits of the model are known. Although removal of the model and observational tritium estimate differences is not completely satisfactory, it appears that existing tritium observations would only weakly constrain a more realistic circulation inversion. The computation, however, does suggest a strategy for future use of transient tracers for constraining circulation models. ¿ American Geophysical Union 1990