Measurements of 234Th (t1/2 = 24.1 days) in dissolved, colloidal, and particulate forms have been made to investigate the role of colloids in reactive metal scavenging in the surface waters of Buzzards Bay, over an annual cycle, and in the shelf and slope waters off New England. At-sea sampling involved prefiltering seawater through 0.2 μm filters followed by cross-flow filtration using a 10,000 nominal molecular weight filter to collect colloidal (10,000 NMW-0.2 μm) and dissolved (<10,000 NMW) phases. Total 234Th activities increase with distance from shore, indicative of enhanced scavenging in the particle-rich nearshore waters. Clearly seen in Buzzards Bay are seasonal changes in total 234Th, with activities ranging from ≈0.7 dpm I−1 in the winter, preceeding a phytoplankton bloom, to ≈0.2 dpm I−1 in the summer. Throughout the annual cycle, 2–16% of total 234Th is colloidal, 22–40% is dissolved, and 45–75% is particulate. In the offshore waters, ≈1% of total 234Th is colloidal, 2–6% is particulate, and 93–98% is dissolved. The 234Th size-distribution exhibits a systematic increase in the association of 234Th with particulate and, to a lesser extent, colloidal matter with increasing suspended particle concentration (Cp). Moreover, a first-order prediction of the fractionation of 234Th between the various size classes is demonstrated using measured solid-solution partition coefficients. Box model calculations indicate a mean residence time of colloidal 234Th with respect to aggregation into particles of 0.3 days in Buzzards Bay, which compares with 2 days for dissolved and 4 days for particulate 234Th. In the offshore surface waters, colloidal and particulate 234Th residence times are ≈0.5 days and 2–3 days respectively, compared with 30–85 days for the dissolved phase. The short and relatively invariant residence time of colloidal 234Th suggests that colloidal aggregation may not be rate-limiting in controlling the scavenging of thorium and, by analogy, other particle-reactive trace metals. An implication of these results is that colloidal 234Th may be tracing a reactive intermediate in the bacterially mediated decomposition of large, rapidly-sinking biogenic aggregates. Using the size-fractionated 234Th data, we demonstrate that Kd values for thorium are invariant with Cp and that scavenging rate constants exhibit a first-order dependence on Cp. Thus, “particle-concentration effects” are negligible for oceanic waters (Cp ≈0.01–1 mg I−1).