Dyshomeostasis of iron metabolism is accounted in the pathophysiological framework of numerous diseases, including cancer and several neurodegenerative conditions. Excessive iron results in free redox-active Fe(II) and can cause devastating effects within the cell like oxidative stress (OS) and death by lipid peroxidation known as ferroptosis (FPT). Therefore, quantitative measurements of ferrous (Fe(II)) and ferric (Fe(III)) iron rather than total Fe-determination is the key for closer insight into these detrimental processes. Since Fe(II)/(III) determinations can be hampered by fast redox-state shifts and low concentrations in relevant samples, like cerebrospinal fluid (CSF), methods should be available that analyze quickly and provide low limits of quantification (LOQ). Capillary electrophoresis (CE) offers the advantage of fast Fe(II)/ Fe(III) separation and works without a stationary phase, which could interfere with the redox balance or cause analyte sticking. CE combined with inductively coupled plasma mass spectrometry (ICP-MS) as a detector offers further improvement of detection sensitivity and selectivity. The presented method uses 20 mM HCl as a background electrolyte and a voltage of +25 kV. Peak shapes and concentration detection limits are improved by conductivity-pH-stacking. For reduction of (56)[ArO](+), ICP-MS was operated in the dynamic reaction cell (DRC) mode with NH3 as a reaction gas. The method achieves a limit of detection (LOD) of 3 mu g/L. Due to stacking, higher injection volumes were possible without hampering separation but improving LOD. Calibrations related to peak area were linear up to 150 mu g/L. Measurement precision was 2.2% (Fe(III)) to 3.5% (Fe(II)). Migration time precision was <3% for both species, determined in 1:2 diluted lysates of human neuroblastoma (SH-SY5Y) cells. Recovery experiments with standard addition revealed accuracy of 97% Fe(III) and 105 % Fe(II). In real-life bio-samples like CSF, migration time can vary according to varying conductivity (i.e., salinity). Thus, peak identification is confirmed by standard addition.