Passive microwave sensors offer the opportunity to retrieve surface SM (SSM) information from the measured brightness temperatures (TB) signals. Nevertheless, these microwave sensors provide individually SSM datasets which may not be consistent worldwide and across time. Therefore, the European Space Agency (ESA) established the Climate Change Initiative (CCI) project with a purpose to merge different observations acquired by several microwave sensors in an attempt to produce the most complete and consistent long-term time series of SSM over 1978-2010 as a first phase [1]. A new innovation in space technology, namely the SMOS (Soil Moisture and Ocean Salinity) SSM datasets has been providing multi-angular microwave TB observations at L-band [2] since 2010. SSM can be retrieved from the SMOS TB observations using several approaches such as a Radiative Transfer based model inversion (e.g. the ESA and CNES operational algorithms [2]), neural networks [3], and statistical regressions. The SMOS satellite being the first mission specifically designed to measure SSM from space, this study -an ESA-funded project- aims at studying the inclusion of SMOS SSM as reference for the long-term SSM datasets. Specifically, investigating the use of physically based multiple-linear regressions [4] to retrieve a global and long-term (e.g. 2003-2014) SSM record based on a combination of passive microwave remote sensing observations from the AMSR-E (2003 - 2011) and SMOS (2010 - 2014) sensors. The overlap measurement period (June 2010-Sept 2011) between AMSR-E and SMOS was used for calibration and validation. The coefficients of the regression equations were calibrated using AMSR-E TB and SMOS Level 3 (SMOSL3, V2.72) SSM (used here as a reference) over the Oct 2010 - Sept 2011 period while the rest of the overlap period (June – Sept 2010) was used for validation. Based on the calibrated coefficients, global SSM maps were computed from the AMSR-E TB observations over the validation period (referred here to as AMSR-reg product). First results showed that the regression approach is very promising as it produces realistic SSM values from the AMSR-E TB product in terms of absolute values. Fig 1 displays a comparison between the temporal mean of the SMOSL3 and AMSR-reg SSM products over the validation period (June – Sept 2010). The global distributions of SSM and its spatial patterns are similar for both products with low values over arid and semi-arid regions, moderate values over India, the Sahel and the Eastern part of Australia. An overall mean of SSM of 0.129 m3/m3 and 0.128 m3/m3 was computed for SMOSL3 and AMSR-reg, respectively. This study is in progress and further investigations will be directed to (i) evaluate the AMSR-reg SSM retrievals against land surface SSM simulations (e.g., MERRA-Land) and in situ observations over the period 2003-2009, at the global and local scales and (ii) investigate the temporal consistency of AMSR-reg (2003-2009) and SMOSL3 (2010-2014) SSM time series. Based on the results, the final goal of this study will be to merge the AMSR-reg (2003-2009) and SMOSL3 (2010-2014) SSM time series to produce a long term and consistent SSM product.