Base flow in the upper San Pedro River at the gaging station (USGS station 09471550) near Tombstone, Arizona, is an important factor in the long-term sustainability of the river's riparian ecosystem. Most base flow occurs during the non-summer months (typically, from November to May), because evapotranspiration (ET) is greater than groundwater discharge to the riparian zone during the growing season and typically causes periods of zero flow in the spring and fall. Streamflow during the summer months occurs only as a result of rainfall and runoff. Using a hydrograph separation technique that partitions streamflow into stormflow and base flow, based on the change in runoff from the previous day, median base flow at the Tombstone gage from 1968 to 2009 (1987 to 1996 data absent) is 4,890 acre-ft/yr. Median base flow for the earlier period of record, 1968 to 1986, is 5,830 acre-ft/yr and for the later period, 1997 to 2009, is 2,880 acre-ft/yr. Base flow in the upper San Pedro River is derived from groundwater discharge to the river from the regional and alluvial aquifer. The regional aquifer is defined as having recharge zones away from the river, primarily at mountain fronts and along ephemeral channels. The alluvial aquifer is recharged mainly from stormflow. Based on environmental isotope data, the composition of base flow in the upper San Pedro River at the gaging station near Tombstone is 74 +/- 10 percent regional groundwater and 26 +/- 10 percent summer storm runoff stored as alluvial groundwater for the 2000 to 2009 period. The volume of base flow in a given year is well explained, using multiple regression, by mean daily flow during the previous October and by rainfall during the months of December and January (R2 = 0.9). This does not suggest that streamflow is composed only of these two sources; rather, these two sources control the degree of saturation of the near-stream alluvial aquifer and, therefore, the amount of winter base-flow infiltration that is possible upstream of the Tombstone gaging station. Because of losing conditions upstream of the Tombstone gage, there is no minimum amount of base flow that would be expected in any given year. The regression equation was used to adjust the measured base flow to account for year-to-year variation in precipitation. Adjusted base flows decreased, independent of climate, from the early period of record to the late period of record. In addition to total base flow, other metrics were considered, including the start and end dates of base flow, the number of days of base flow, the 25th percentile mean daily flow, and the number of days of zero flow. Each of these showed a decline in base flow between the early period of record and the late period. The available evidence to evaluate this decrease - hydraulic gradients in the alluvial and regional aquifers and a 10-yr record of streamflow environmental isotope samples - indicates that no reduction in groundwater discharge has occurred over this period of record. Continued regional groundwater pumping will, however, eventually lead to a decline in the contribution of regional groundwater to base flow.