The distribution of 234Th, 230Th, and 228Th between dissolved and particulate forms was determined in 17 seawater samples from the Guatemala and Panama basins. Sampling was carried out in situ with battery-powered, submersible pumping systems in which the seawater first passed through a Nuclepore filter (1.0-μm pore size) and then through a cartridge packed with Nitex netting that was impregnated with MnO2 to scavenge the dissolved Th isotopes. Natural 234Th was used as the tracer for monitoring the efficiency of scavenging. For all three isotopes, most of the activity was found in the dissolved form. On the average 4% of the 234Th, 15% of the 228Th, and 17% of the 230Th occurred in the particulate form, though the percentages were found to be strongly dependent on particle concentration. These distributions are not consistent with chemical scavenging models that assume irreversible uptake of Th on particle surfaces. The results can be explained, however, if continuous exchange of Th isotopes between seawater and the particle surfaces is assumed. Vertical profiles of both particulate and dissolved 230Th show increasing concentrations with depth, as required by the assumption of reversible exchange. Some of the dissolved 230Th profiles, however, show a reversal of this trend near the bottom, indicating accelerated scavenging near the water/sediment interface. Kinetics of both adsorption and desorption can be examined if at least two Th isotopes are measured in the same samples. Results show that reaction times are short (a few months) compared to the residence time of suspended matter in the deep ocean (several years), indicating that particles suspended in the deep sea are close to equilibrium with respect to exchange of metals at their surfaces.