The reaction of lithium with SnFe is studied using in situ Fe-57 and Sn-119 Mossbauer spectroscopy and correlated to results measured using X-ray diffraction. During the first discharge of Li/SnFe cells, the cell reaction is approximately 2/3 4.4Li + SnFe --> 2/3 Li4.4Sn + 2/3 Fe + 1/3 SnFe. The reaction does not go to completion because the iron which is displaced forms a 'skin' on the remaining SnFe, preventing complete reaction. During the first charge (removing Li from the Li-Sn alloys) there are dramatic changes in the Fe-57 and Sn-119 Mossbauer spectra which indicate that many of the 'liberated' Sn atoms 'back react' with Fe to form ferromagnetic grains of Sn-Fe alloys. We believe these grains have the bcc Fe structure with some substitutional Sn. The original SnFe is not reformed, although the 'skinned' original SnFe remains. During the next discharge, these Sn-Fe alloy grains react with Li to form Li4.4Sn and Fe again. As the cells are consecutively charged and discharged, the hyperfine magnetic field in the bcc Sn-Fe alloy formed during charge increases to near the value in iron. We take this as evidence that the tin content of this phase gets smaller and smaller with cycle number, implying that there must also be tin grains in the charged electrode. As the cycle number increases, both X-ray diffraction and Mossbauer spectroscopy show that the amount of unreacted SnFe in the electrode gets smaller and smaller, suggesting that the Fe 'skin' on the unreacted particles is slowly breached. Once the cell capacity decays to about 50% of its initial value, X-ray diffraction and Mossbauer spectroscopy of the charged electrodes (as much Li as possible removed) show electrically disconnected Li4.4Sn, and bcc Sn-Fe alloy which has very small tin content. Cell failure occurs because the grain size of the Li-Sn alloy and Sn-Fe regions becomes larger with cycle number leading to electrical disconnection.