Two methods, one a surface-water method and the second a ground-water method, were used to determine infiltration and percolation rates along a 2.5-kilometer reach of the Santa Fe River near La Bajada, New Mexico. The surface-water method uses streamflow measurements and their differences along a stream reach, streamflow-loss rates, stream surface area, and evaporation rates to determine infiltration rates. The ground-water method uses heat as a tracer to monitor percolation through shallow streambed sediments. Data collection began in October 1996 and continued through December 1997. During that period the stream reach was instrumented with three streamflow gages, and temperature profiles were monitored from the stream-sediment interface to about 3 meters below the streambed at four sites along the reach. Infiltration is the downward flow of water through the stream- sediment interface. Infiltration rates ranged from 92 to 267 millimeters per day for an intense measurement period during June 26- 28, 1997, and from 69 to 256 millimeters per day during September 27-October 6, 1997. Investigators calculated infiltration rates from streamflow loss, stream surface-area measurements, and evaporation-rate estimates. Infiltration rates may be affected by unmeasured irrigation-return flow in the study reach. Although the amount of irrigation-return flow was none to very small, it may result in underestimation of infiltration rates. The infiltration portion of streamflow loss was much greater than the evaporation portion. Infiltration accounted for about 92 to 98 percent of streamflow loss. Evaporation-rate estimates ranged from 3.4 to 7.6 millimeters per day based on pan-evaporation data collected at Cochiti Dam, New Mexico, and accounted for about 2 to 8 percent of streamflow loss. Percolation is the movement of water through saturated or unsaturated sediments below the stream-sediment interface. Percolation rates ranged from 40 to 109 millimeters per day during June 26-28, 1997. Percolation rates were not calculated for the September 27-October 6, 1997, period because a late summer flood removed the temperature sensors from the streambed. Investigators used a heat-and-water flow model, VS2DH (variably saturated, two- dimensional heat), to calculate near-surface streambed infiltration and percolation rates from temperatures measured in the stream and streambed. Near the stream-sediment interface, infiltration and percolation rates are comparable. Comparison of infiltration and percolation rates showed that infiltration rates were greater than percolation rates. The method used to calculate infiltration rates accounted for net loss or gain over the entire stream reach, whereas the method used to calculate percolation was dependent on point measurements and, as applied in this study, neglected the nonvertical component of heat and water fluxes. In general, using the ground-water method was less labor intensive than making a series of streamflow measurements and relied on temperature, an easily measured property. The ground-water method also eliminated the difficulty of measuring or estimating evaporation from the water surface and was therefore more direct. Both methods are difficult to use during periods of flood flow. The ground-water method has problems with the thermocouple-wire temperature sensors washing out during flood events. The surface- water method often cannot be used because of safety concerns for personnel making wading streamflow measurements.