[1] Large volume (10²-10³ L) seawater samples are routinely processed to investigate the partitioning of particle reactive radionuclides and Ra between solution and size-fractionated suspended particulate matter. One of the most frequently used methods to preconcentrate these nuclides from such large volumes involves extraction onto three filter cartridges (a prefilter for particulate species and two MnO₂-coated filters for dissolved species) connected in series. This method assumes that the extraction efficiency is uniform for both MnO₂-coated cartridges, that no dissolved species are removed by the prefilter, and that any adsorbed radionuclides are not desorbed from the MnO₂-coated cartridges during filtration. In this study, we utilized ²³⁴Th-spiked coastal seawater and deionized water to address the removal of dissolved Th onto prefilters and MnO₂-coated filter cartridges. Experimental results provide the first data that indicate (1) a small fraction of dissolved Th (<6%) can be removed by the prefilter cartridge; (2) a small fraction of dissolved Th (<5%) retained by the MnO₂ surface can also be desorbed, which undermines the assumption of uniform extraction efficiency for Th; and (3) the absolute and relative extraction efficiencies can vary widely. These experiments provide insight on the variability of the extraction efficiency of MnO ₂-coated filter cartridges by comparing the relative and absolute efficiencies and recommend the use of a constant efficiency on the combined activity from two filter cartridges connected in series for future studies of dissolved ²³⁴Th and other radionuclides in natural waters using sequential filtration/extraction methods. ?? 2009 by the American Geophysical Union.