Generalized models of thorium and particle cycling, data from Station P, and an inversion technique are used to obtain rate estimates of important biological and chemical transformations occurring in the water column. We first verify the inversion technique using an idealized data set generated by a finite difference model, and then apply the inversion technique to data from Station P. With the Station P data, predicted rate constants for adsorption and release of thorium between the dissolved and small particle phases are consistent with the results from other workers. The predicted rate constants for the interaction between small and large particles are smaller than previous estimates. The predicted concentration of large rapidly sinking particles is greater than the concentration of suspended non-sinking particles, whereas the reverse is usually assumed to be the case. The calculated sinking rate for the large particles is 20 m d-1. This sinking rate is an order of magnitude smaller than the large particle sinking rate inferred from sediment trap mass fluxes at two levels in the water column. The reason we predict a high large particle concentration and slow settling velocity has not been uniquely determined. Possible modifications of the current model that could help to reconcile the differences between observations and model predictions include: 1) two classes of rapidly sinking particles or rate constants that change with depth, 2) direct interactions between the large particle and dissolved phases, and 3) incorporation of a continuous distribution of particle size and settling velocity. ¿ American Geophysical Union 1990