The quantification of river-aquifer interaction is critical to the conjunctive management of surface water and groundwater, in particular in the arid and semiarid environment with much higher potential evapotranspiration than precipitation. A variety of natural tracer methods are available to quantify river-aquifer interaction at different scales. These methods however have only been tested in rivers with relatively low flow rates (mostly less than 5 m3 s-1). In this study, several natural tracers including heat, radon-222 and electrical conductivity were measured both on vertical riverbed profiles and on longitudinal river samples to quantify river-aquifer exchange flux at both point and regional scales in the Heihe River (northwest China; flow rate 63 m3 s-1). Results show that the radon-222 profile method can estimate a narrower range of point-scale flux than the temperature profile method. In particular, three vertical radon-222 profiles failed to estimate the upper bounds of plausible flux ranges. Results also show that when quantifying regional-scale river-aquifer exchange flux, the river chemistry method constrained the flux (5.20-10.39 m2 d-1) better than the river temperature method (-100 to 100 m2 d-1). The river chemistry method also identified spatial variability of flux, whereas the river temperature method did not have sufficient resolution. Overall, for quantifying river-aquifer exchange flux in a large river, both the temperature profile method and the radon-222 profile method provide useful complementary information at the point scale to complement each other, whereas the river chemistry method is recommended over the river temperature method at the regional scale.