Lithium-ion batteries require higher energy densities to meet with a broad acceptance in the fields of electric vehicles and grid storage solutions. LiNi0.5Mn1.5O4 (LNMO) can fulfill this goal due to its high operating voltage. Cycling of LNMO is known to be stable vs. lithium metal anode. Cycling in an LNMO/graphite configuration leads to severe capacity fade. Ti-doped LNMO (LNMTO)/graphite cells experience a lower, but still strong loss of capacity. In order to understand capacity fade, cycling tests of LNMTO vs. graphite and vs. lithium metal were carried out and additionally, three electrode tests were performed. Both cell configurations showed similar Coulombic efficiencies correlating with the applied C-rate. Experimental data and mathematical modeling indicated that loss of active lithium with a constant reaction rate of (3.76 +/- 0.46) . 10(-8) mol Li h(-1) is responsible for capacity fade in LNMTO/graphite cells and that no degradation of the active material occurs. It was concluded that lithium loss also occurs when lithium metal anodes are used. Here, the lithium metal anode can compensate for lithium consumption, as a result of which the capacity is not influenced. Further support for lithium consumption is given by a three-electrode cell with a lithiated graphite anode. The lithium in the graphite anode can compensate the lithium loss for 120 cycles. During this time, the cell experienced hardly any capacity fade and the voltage profile was similar to that of a cell with LNMTO/Li configuration. (C) 2014 Elsevier B.V. All rights reserved.