This paper focuses on the determination of experimental uncertainties based on an error analysis of experimental three-phase equilibria data. An objective function is minimized such that all material balance and physical constraints are satisfied. This function is defined in terms of the optimized phase compositions and variable experimental uncertainties of the experiments. To test the procedure, we used three-phase compositions measured in our laboratory from a quaternary system of propane (C-3), n-pentane (C-5), n-octane (C-8), and water (H2O) at three different pressures and temperatures. We also analysed published three-phase equilibria data for a quaternary system of methane (CH4), carbon dioxide (CO2), hydrogen sulfide (H2S), and water (H2O). Unbiased maximum experimental uncertainties were estimated for all these cases. These uncertainties ranged from 0.0004 to 0.01 mole fraction units. With the largest experimental uncertainty, the phase molar fractions determined from raw experimental data could be up to 9% different from those determined with optimized, self-consistent compositions.