This paper presents a transformation of a SAT problem into another SAT problem which, in turn, can be directly transformed into a binary CSP instance. This transformation is based on a local model enumeration procedure which can be controlled to get an overall polynomial transformation. A method to reduce the size of the converted formula is proposed. We also show that maintaining k-consistency on the CSP problem is equivalent to using unit propagation on a compiled form of the equivalent SAT problem. This transformation draws another natural bridge between the SAT and the CSP problems which hopefully, should benefit both communities.