This work presents a metrological insight into correctness of the results obtained in standard addition voltammetric determination of the selected antioxidants on a carbon-fiber microelectrode. The key issue is taking into account the uncertainty in both axes during the process of estimation of the calibration curve coefficients and results of analysis. It has been shown that in many cases, taking into account more sources of uncertainty leads to more accurate results than when using typical linear regression. Additionally, the level of the uncertainty is also lower. The problem was demonstrated on the examples of determination of micromolar concentrations of caffeic, syringic and vanillic acids in laboratory samples and artificial wine, which was regarded as a model matrix. It was observed that sensitivity for caffeic acid was ca. 8-9 times higher than for syringic acid and vanillic acid. In all experiments correlation coefficient was greater than 0.996, while recovery was between 91.5 and 117.5%. The obtained result of determination of 2 mu M caffeic acid was 2.33 +/- 0.10 mu M, while considering uncertainties in both axes was 2.31 +/- 0.04 mu M. Similarly, for 5 mu M syringic acid we obtained 5.91 +/- 0.47 mu M and 5.52 +/- 0.37 mu M, while for 10 mu M vanillic acid it was 9.21 +/- 1.05 mu M and 9.29 +/- 0.80 mu M. (C) 2016 Elsevier B.V. All rights reserved.