Estimation of submarine groundwater discharge (SGD) to semi-enclosed basins by Ra isotope mass balance is herein assessed. We evaluate 224 Ra, 226 Ra, and 228 Ra distributions in surface and bottom waters of Long Island Sound (CT-NY, United States) collected during spring 2009 and summer 2010. Surface water and bottom water Ra activities display an apparent seasonality, with greater activities during the summer. Long-lived Ra isotope mass balances are highly sensitive to boundary fluxes (water flux and Ra activity). Variation (50%) in the 224 Ra, 226 Ra, and 228 Ra offshore seawater activity results in a 63-74% change in the basin-wide 226 Ra SGD flux and a 58-60% change in the 228 Ra SGD flux, but only a 4-9% change in the 224 Ra SGD flux. This highlights the need to accurately constrain long-lived Ra activities in the inflowing and outflowing water, as well as water fluxes across boundaries. Short-lived Ra isotope mass balances are sensitive to internal Ra fluxes, including desorption from resuspended particles and inputs from sediment diffusion and bioturbation. A 50% increase in the sediment diffusive flux of 224 Ra, 226 Ra, and 228 Ra results in a ∼30% decrease in the 224 Ra SGD flux, but only a ∼6-10% decrease in the 226 Ra and 228 Ra SGD flux. When boundary mixing is uncertain, 224 Ra is the preferred tracer of SGD if sediment contributions are adequately constrained. When boundary mixing is well-constrained, 226 Ra and 228 Ra are the preferred tracers of SGD, as sediment contributions become less important. A three-dimensional numerical model is used to constrain boundary mixing in Long Island Sound (LIS), with mean SGD fluxes of 1.2 ± 0.9 × 10 13 L y −1 during spring 2009 and 3.3 ± 0.7 × 10 13 L y −1 during summer 2010. The SGD flux to LIS during summer 2010 was one order of magnitude greater than the freshwater inflow from the Connecticut River. The maximum marine SGD-driven N flux is 14 ± 11 × 10 8 mol N y −1 and rivals the N load of the Connecticut River.