Estimation of the average annual rate of ground-water recharge to sand and gravel aquifers using elevated tritium concentrations in ground water is an alternative to traditional steady-state and water-balance recharge-rate methods. The concept of the tritium tracer method is that the average annual rate of ground-water recharge over a period of time can be calculated from the depth of the peak tritium concentration in the aquifer. Assuming that ground-water flow is vertically downward and that aquifer properties are reasonably homogeneous, and knowing the date of maximum tritium concentration in precipitation and the current depth to the tritium peak from the water table, the average recharge rate can be calculated. The method, which is a direct-measurement technique, was applied at two sites on Nantucket Island, Massachusetts. At site 1, the average annual recharge rate between 1964 and 1983 was 26.1 inches per year, or 68 percent of the average annual precipitation, and the estimated uncertainty is ?15 percent. At site 2, the multilevel water samplers were not constructed deep enough to determine the peak concentration of tritium in ground water. The tritium profile at site 2 resembles the upper part of the tritium profile at site 1 and indicates that the average recharge rate was at least 16 .7 inches per year, or at least 44 percent of the average annual precipitation. The Nantucket tritium recharge rates clearly are higher than rates determined elsewhere in southeastern Massachusetts using the tritium, water-table-fluctuation, and water-balance (Thornthwaite) methods, regardless of the method or the area. Because the recharge potential on Nantucket is so high (runoff is only 2 percent of the total water balance), the tritium recharge rates probably represent the effective upper limit for ground-water recharge in this region. The recharge-rate values used by Guswa and LeBlanc (1985) and LeBlanc (1984) in their ground-water-flow computer models of Cape Cod are 20 to 30 percent lower than this upper limit. The accuracy of the tritium method is dependent on two key factors: the accuracy of the effective-porosity data, and the sampling interval used at the site. For some sites, the need for recharge-rate data may require a determination as statistically accurate as that which can be provided by the tritium method. However, the tritium method is more costly and more time consuming than the other methods because numerous wells must be drilled and installed and because many water samples must be analyzed for tritium, to a very small level of analytical detection. For many sites, a less accurate, less expensive, and faster method of recharge-rate determination might be more satisfactory . The factor that most seriously limits the usefulness of the tritium tracer method is the current depth of the tritium peak. Water with peak concentrations of tritium entered the ground more than 20 years ago, and, according to the Nantucket data, that water now is more than 100 feet below the land surface. This suggests that the tracer method will work only in sand and gravel aquifers that are exceedingly thick by New England standards. Conversely, the results suggest that the method may work in areas where saturated thicknesses are less than 100 feet and the rate of vertical ground-water movement is relatively slow, such as in till and in silt- and clay-rich sand and gravel deposits.