Seismic bright spots are commonly interpreted to mark fluid concentrations, but their nature (melt or aqueous) is usually inferred only from circumstantial evidence of the geologic setting. A band of bright spot reflections has been imaged by Project INDEPTH (International Deep Profiling of Tibet and the Himalayas) at about 15 km depth along 150 km of the northern Yadong-Gulu rift, southern Tibet. We use INDEPTH three-component wide-angle seismic data to measure seismic velocities at the bright spot reflector, and theoretical rock physics bounds to constrain the nature of the fluids. Merging of data from multiple bright spots allows us to use a one-dimensional approximation. Travel time modeling yields average P and S velocities for the upper crust above the bright spots of 5.3¿0.2 and 3.2¿0.2 km s-1, respectively. Reflection-amplitude variation with offset (AVO) modeling constrains the P and S velocities of the bright spots to 3.0¿0.8 and 1.6¿0.8 km s-1, respectively. Multiple modeling procedures suggest these velocities are not model dependent. Our results imply that of the order of 10% volume of free aqueous fluids in the Tibetan middle crust produces the observed bright spot reflections. The presence of relatively large quantities of free aqueous fluids, presumably mostly saline supercritical H2O, does not preclude the presence of melt but does constrain the maximum temperature at the bright spots to the wet granite solidus (about 650 ¿C) and thus the maximum surface heat flow to ≤110 mW m-2. The observed bright spots can alternatively be explained as a result of transient flow of aqueous fluids through a lower temperature and lower heat flow southern Tibetan crust. ¿ 1999 American Geophysical Union