The concentration (C) of dissolved 238 U, 234 U, 232 Th and 230 Th in fresh and brackish waters from the Baltic Sea were determined using TIMS. The brackish waters range in salinity from that of sea water (SW) to 2.5‰. C 238U in oxygen-saturated, surface waters is well correlated with salinity and shows quasi-conservative behavior, as does Sr. Samples from the redox water interface show depletion in C 238U , demonstrating that dissolved U is being removed by FeMn oxyhydroxides. From a simple mixing relationship for the brackish water,δ 234 U * = 1000‰ was calculated for the fresh water source in the northern Baltic. A study of the Kalixälven River over an annual cycle yields highδ 234 U during spring and summer discharge and lower values during fall and winter, showing that different sources contribute to the U load in the river during different seasons. C 232Th and C 230Th in river water are governed by the discharge, reflecting the importance of the increased abundance of small particles ( < 0.45 μm) for the 232 Th 230 Th load at high discharge. 232 Th/ 238 U in river water is about 40 times less than in detrital material. In the brackish water, C232 Th drops 2 orders of magnitude in the low salinity region ( < 5‰), reaching a value close to that of sea water at a salinity of 7.5‰. Almost all of the riverine 232 Th must be deposited in the low-salinity regions of the estuary.The 230 Th/ 232 Th in river waters is about twice the equilibrium value for 232 Th/ 238 U (3.8). In the brackish waters, 230 Th/ 232 Th is greater by a factor of 10–100 than both river water and SW. The big increase in 230 Th/ 232 Th in the Baltic Sea waters over the riverine input indicates that the Th isotopes enter the estuary as a mixture of two carrier phases. We infer that about 96% of 232 Th in river water is carried by detrital particles, whereas the other phase (solution, colloidal) has a much higher 232 Th/ 232 Th. Entering the estuary, the detrital particles sediment out rapidly, whereas the non-detrital phase is removed more slowly, causing a marked increase in 230 Th/ 232 Th in the brackish water. In SW, 230 Th/ 232 Th is closer to river input and detrital material than in brackish water. We conclude that in the deep sea, 232 Th is almost exclusively dominated by windblown dust and can be used to monitor dust flux. The 230 Th excess in Baltic rivers is produced in U-rich, 232 Th-poor peatlands and trapped in authigenic particles and transported with the particles. Time scales for producing the 230 Th excess are ∼ 2000–8000 yr. This is younger than, but comparable to, the time of the latest deglaciation, which ended some 9000 yr ago when the mires were forming. These results have implications for the possible mobility of actinides stored in repositories.