The purpose of this study is to understand the behaviour of the actinides in groundwater by combining data on the distribution of both longlived and short-lived nuclides of the U and Th decay series in an aquifer with a model for the nuclide transport along aquifer flow lines. The groundwater concentrations of the short-lived nuclides, such as 222Rn, 224Ra and 234Th, provide information only on local aquifer conditions, while the concentrations of the long-lived nuclides reflect geochemical processes along extended flow paths. In an important study, Krishnaswami et al. (1982) calculated sorption rate constants for many of the decay series nuclides in an aquifer using a batch model. However, no study yet has combined the data provided by both long and short-lived radionuclides to study the transport mechanisms along a flow line. Furthermore, although colloidal phases have been found to be important carriers of U and Th in numerous natural aqueous systems, very few investigations have been conducted on their importance in the transport of naturally-occurring actinides in groundwater. The aquifer investigated here is the sandy unconfined Upper Glacial Pleistocene aquifer situated in Long Island, NY. Extensive information regarding the geology and the hydrology of the aquifer is available and numerous high quality and well-documented wells are available for sample collection. This aquifer is underlain by a discontinuous layer of impermeable clay, below which is a sandy confined aquifer, the Magothy, where reducing conditions prevail. Radionuclide data from several wells in this aquifer have been reported by Copenhaver et al. (1993). Samples from the Upper Glacial aquifer have been taken along two flow lines at three different depths and filtered through 0.45 gm filter cartridges. In addition, cross-flow ultrafiltration has been used to separate colloids >10 kD from ultrafiltered water that represents the dissolved load. This technique, described in previous papers, has proven to be successful in studying the association of uranium and other elements with colloids (e.g. Porcelli et al., 1997). The concentrations of the long-lived radionuclides :3SU, 234U and 23eTh have been measured by thermal ionization mass spectrometry and the short-lived radiogenic daughters 222Rn, 226Ra, 228Ra, 224Ra, 234Th by counting methods. In addition, the water samples have been analysed for pH and conductivity in the field, and for concentrations of major cations, anions and trace elements. Water samples from the Upper Glacial aquifer have pH values of 5.4 to 6.5 and TDS (Total Dissolved Solid) of 20 to 100 mg/l. The high Mn concentration in a deep sample of the shallow aquifer likely indicates a difference in redox conditions. The Magothy aquifer has TDS o f 50 mg/1, comparable to that of the unconfined aquifer, but much higher Fe and Mn concentrations. The 222Rn activities generally full within a restricted range throughout the Upper Glacial aquifer of 65 to 170 dpm/kg, with two peak values of 3 36 and 522 dprn/kg. The amount of Rn in the water represents -4 % of the Rn produced in the rock by decay. Assuming that the aquifer rock consists of solid spherical grains with uniform distribution of Ra, this amount of Rn can only be supplied by grains of ~ 0.1 ~tm in size and therefore cannot be produced by the aquifer which average grain size is ~1 mm. This has been observed previously (e.g. Copenhaver et al., 1993). High Rn concentrations have been observed and lead to the requirement that Rn is circulating through nanopores or that the source of Rn is on the surface. We observe that the dissolved Rn concentration is generally greater where the dissolved 226Ra concentration is lower, suggesting that higher Rn concentrations in the water are due to increased amounts of 226Ra on aquifer surfaces due to deposition of dissolved 226Ra. Therefore we infer that Rn concentrations here are controlled by the distribution of Ra on surfaces rather than by enhanced release from