The main goal of this paper is the introduction of a predictive model to calculate the remaining run-time of a battery. It is not meant to increase precision or compete with other methods but to be a reliable model by using a new formulation, more representative of the active material used zones. This paper defines two different states of charge (SOC) based on the concepts of standard and available capacities, named SOCs and SOCav, respectively. Concepts such as uncharged and undischarged capacities and loss charge efficiency are included as well. With these new definitions, an estimation algorithm is proposed. It is implemented as a runtime SOC estimation model based on an ampere-hour counting electrical model. The characterization of SOC model parameters of an 11 Ah Ni-Cd battery is also presented. These tests combine non-standard charge or discharge processes with end-of-charge/discharge detection methods that avoid overcharge or over discharge. Four case studies are carried out: two on a single battery cell and the other two on a 210 battery stack. The results show a good performance, with an approximate improvement of the estimation of 5%. They also show the importance of differentiating standard and available SOC in order to calculate the available capacity. Published by AIP Publishing.