Abstract Application of spaceborne reflectance measurements of remote sensing data from the cyclone global navigation satellite system(CYGNSS) and soil moisture active passive(SMAP) soil moisture products, two methods for surface soil moisture retrieval from spaceborne global navigation satellites system reflectometry (GNSS-R) are compared and evaluated. The results show that, the simple linear regression modeling between calibrated gridded effective reflectivity and reference SMAP soil moisture on each aggregated grid pixel is superior to the integrated modeling over global. An inversion model with significant time-domain stability can be produced with 1 a of the training data. On a five-month testing dataset, the RMSE of the estimated soil moisture from the pixel-by-pixel retrieval model can reach 0.056 cm3/cm3, and the correlation between estimated soil moisture and reference soil moisture is 0.9.