We report a promising approach for on-line estimating the capacity of a Li-ion battery based on half-cell open circuit voltages(OCV). Capacity estimation of a rechargeable battery is especially challenging for those on-line applications where a full charge or discharge of the battery is not allowed. Additionally, conventional OCV based capacity estimation methods do not work well for those batteries whose OCV similar to SOC curves change substantially during their life. Based on the recent experimental findings we assume that the capacity loss is solely due to the loss of active Li-ions inventory, which enables us to derive the battery's OCV similar to SOC curves at different ages using its half-cell OCV similar to SOC curves at the beginning of life. The good agreement between the predicted and the measured curves at different cell ages (capacity states) justifies the accuracy of this derivation. The generated OCV similar to SOC curves can be used as look-up table (LUT)s during battery use to determine the cell capacity in real time. We demonstrated the on-line battery capacity estimations with a discharge pulse on aged commercial Li-ion cells and achieved accuracies within 1% error. The half-cell OCV based approach offers a new and effective technique for on-line capacity estimation of Li-ion batteries. (C) 2014 The Electrochemical Society. All rights reserved.