A new process decomposed calculation method is developed to compare the cycle based charge, discharge, net, and overall energy efficiencies of lithium-ion batteries. Multi-cycle measurements for both constant current (CC) and constant current to constant voltage (CC-CV) charge models have been performed. Unlike most conventional efficiency calculation methods with one mean open-circuit voltage (OCV) curve, two OCV curves are calculated separately for the charge and discharge processes. These two OCV curves help to clarify the intra-cycle charge, discharge, net, and overall energy efficiencies. The relationships of efficiencies versus state of charge, state of quantity, and scaled stresses are demonstrated. Efficiency degradation patterns versus cycle numbers and scaled stresses are also illustrated with the artificial neural network (ANN) prediction method. The decaying ratios of the overall efficiencies are about 2% and 0.3% in the first 30 cycles, for CC and CC-CV, respectively. Hence, efficiencies of the CC-CV model are more stable compared with the CC model, which are shown by both experimental and ANN prediction results.