Developing a more thorough understanding of water and chemical budgets in wetlands depends in part on our ability to quantify time-varying interactions between ground water and surface water. We used a combined water and solute mass balance approach to estimate time-varying ground-water discharge and recharge in the Everglades Nutrient Removal project (ENR), a relatively large constructed wetland (1544 hectare) built for removing nutrients from agricultural drainage in the northern Everglades in South Florida, USA. Over a 4-year period (1994 through 1998), ground-water recharge averaged 13.4 hectare-meter per day (ha-m/day) or 0.9 cm/day, which is approximately 31% of surface water pumped into the ENR for treatment. In contrast, ground-water discharge was much smaller (1.4 ha-m/day, or 0.09 cm/day, or 2.8% of water input to ENR for treatment). Using a water-balance approach alone only allowed net ground-water exchange (discharge − recharge) to be estimated (−12 ± 2.4 ha-m/day). Discharge and recharge were individually determined by combining a chloride mass balance with the water balance. For a variety of reasons, the ground-water discharge estimated by the combined mass balance approach was not reliable (1.4 ± 37 ha-m/day). As a result, ground-water interactions could only be reliably estimated by comparing the mass-balance results with other independent approaches, including direct seepage-meter measurements and previous estimates using ground-water modeling. All three independent approaches provided similar estimates of average ground-water recharge, ranging from 13 to 14 ha-m/day. There was also relatively good agreement between ground-water discharge estimates for the mass balance and seepage meter methods, 1.4 and 0.9 ha-m/day, respectively. However, ground-water-flow modeling provided an average discharge estimate that was approximately a factor of four higher (5.4 ha-m/day) than the other two methods. Our study developed an initial understanding of how the design and operation of the ENR increases interactions between ground water and surface water. A considerable portion of recharged ground water (73%) was collected and returned to the ENR by a seepage canal. Additional recharge that was not captured by the seepage canal only occurred when pumped inflow rates to ENR (and ENR water levels) were relatively high. Management of surface water in the northern Everglades therefore clearly has the potential to increase interactions with ground water.