The monsoon freshwater and wind forcing drive high Sea Surface Salinity (SSS) contrasts and variability (up to 10 pss range) in the Bay of Bengal (BoB), with important consequences for upper ocean mixing and air-sea interactions. Synoptic SSS maps did only become available with the advent of L-band radiometers in 2010, due to insufficient prior in situ data coverage. Here, we build tools aiming at reconstructing the monthly BoB SSS at ¼° resolution since 2002 from AMSR-E radiometer data. The C-band low sensitivity to SSS requires a very careful processing. Taking the X- minus C- bands signals reduce the impact of Sea Surface Temperature (SST) and wind on brightness temperatures. It was however further necessary to train the algorithm with SSS data from L-Band radiometers to remove residual surface winds, SST, and atmospheric water contents signals. We also found that a separate treatment of the ascending and descending passes was necessary, as well as a proper data screening to minimize contamination by land signals. The resulting SSS product reproduces the broad BoB climato-logical SSS, and has a 0.66 correlation, 1.08 pss rms-difference to co-located in situ surface salinity from Array for Real-time Geostrophic Oceanography (ARGO) floats. Comparisons with ocean re-analyses in two SSS interannual variability hotspots indicate poor performance in the Northern BoB, but some skill along the East coast of India. Our results provide a proof of concept for reconstructing the BoB SSS from AMSR-E data, and we discuss possible future improvements of the data processing to further reduce the impact of spurious signals.