Tin oxide composite glasses represent a new class of material for the anode of Li-ion cells. Using results of experiments on Li/Sn2BPO6 and Li/SnO2 cells, we identify those factors which are responsible for good charge-discharge capacity retention. First, the grains (those regions which diffract coherently) which make up the particles of the material should be as small as possible. Then, regions of tin which form are kept small and two-phase coexistence regions between bulk Li-Sn alloys of different composition do not occur. The Sn2BPO6 glass represents the smallest grains possible. Second, the particles themselves should be small so that they can each be well contacted by carbon black during electrode manufacture. Third, the voltage range of cycling must be selected so that the tin atoms do not aggregate into large regions which grow in size. This aggregation is evidenced by the growth of peaks in the differential capacity vs. voltage as a function of cycle number. The peaks represent-the coexistence between bulk Li-Sn alloy phases which have substantially different volumes. The coexistence is thought to cause fracturing and loss of contact between the grains. Therefore, materials with small particles, small grains, and smooth sloping voltage profiles which do not change with cycle number (as indicated by a stable differential capacity) give the best cycling performance. The selection of the voltage limits for cycling strongly influences the stability of the voltage profile (as illustrated here), so this must be done with much care.