The growing demand for electric vehicles and renewable energysources has increased the need for safe, reliable, and cost-effectiveenergy-storage systems, many of which include batteries. Thereliability and efficiency of these battery-based systems can besignificantly improved using intelligent energy-management systemsthat effectively indicate battery health in real time. On-linemonitoring can be difficult, however, because batteries are non-linearand time-varying systems whosecharacteristics depend on temperature, usage history, and other factors. The key metrics of interest in a battery are its remaining capacity and health. Most of the current methods require off-line measurement, and even the available on-line methods are only good in laboratory conditions. This thesis provides an enhanced streamlined framework for on-line monitoring. In this methodology, a non-intrusive test signalis superimposed upon a battery load which causes transient dynamics inside the battery. The resulting voltage and current are used as test data and the estimation is done intwo parts. First, a non-linear least-squares routine is used to estimate the electricalparameters of a battery model. Second, a state-estimation algorithm is used to estimate theopen-circuit voltage. Experimental results obtained at consistenttemperatures demonstrate that the open-circuit voltage and parameter values together can combine toprovide capacity and health measurements. This approach requires minimal hardware and could form the basis for a robust on-line monitoring system.