Natural samples were run in three parallel flow-through cells in the laboratory in order to study the difficulties of obtaining stable redox potential readings, especially in gneiss groundwater and mine water with little redox buffer capacity. Redox potentials were recorded every 2 min for up to 5 days. Measured redox potentials were compared to means of partial potentials, modeled based on species-selective determinations of the most predominant redox-sensitive elements iron and nitrogen. Redox potentials stabilized reproducibly within minutes for a synthetic redox buffer solution and two well-buffered acid mine water samples at pH 2.3 and 3.5. For waters in redox disequilibrium, species-selective analytics might still be the better alternative compared to measuring the redox potential as summary parameter for modeling species distribution. In cases where the redox potential is the only readily available parameter, filtration and possibly acidification is recommended for quicker stabilization of redox potential readings and less deviations between measured and modeled values. A minimum of 2% was achieved in the samples investigated, however, often after measuring significantly longer than the previously suggested 30 min. Final stabilization could take up to several hours in waters with low buffer capacity.