Radon is useful as a tracer of certain geophysical processes in marine and aquatic environments. Recent applications include detection of groundwater discharges into surface waters and assessment of air/sea gas piston velocities. Much of the research performed in the past decade has relied on continuous measurements made in the field using a radon stripping unit connected to a radon-in-air detection system. This approach assumes that chemical equilibrium is attained between the water and gas phases and that the resulting air activity can be multiplied by a partition coefficient to obtain the corresponding radon-in-water activity. We report here the results of a series of laboratory experiments that describes the dependence of the partition coefficient upon both water temperature and salinity. Our results show that the temperature dependence for freshwater closely matches results that were previously available. The salinity effect, however, has largely been ignored and our results show that this can result in an overestimation of radon concentrations, especially in cooler, more saline waters. Related overestimates in typical situations range between 10 (warmer less saline waters) and 20% (cooler, more saline waters).