Multi-walled carbon nanotubes (MWCNTs) were used to modify the surface of a glassy carbon electrode (GCE) to enhance its electroactivity. Nafion served to fix the MWCNTs on the GCE surface. The analytical performance of the MWCNT-GCE electrode was assessed using square-wave voltammetry (SWV) and ascorbic acid (AA) as the model analyte. The electrode was generally prepared fresh daily and 0.1 M acetate buffer (pH 3.7) served as the supporting electrolyte in all SWV experiments. The two-electron oxidation of AA at the MWCNT-GCE electrode showed a peak potential at approximately 315 mV, which was 80 mV lower than that at the bare GCE indicating a more energetically favoured oxidation process. The corresponding peak current was about 3-fold higher than the response at a bare GCE, which enhances the AA detection. The peak currents increased with increasing AA concentration (0.25-5 mmol/L) displaying good linearity (R2 = 0.994) as well as good reproducibility of peak currents for replicate measurements (3% RSD). The detection limit (based on 3xsd) was 1 mmol L-1 AA The MWCNT-GCE electrode was applied to the determination of ascorbic acid in four brands of commercial orange juice products. The measured AA content ranged at 0.435-0.469 mmol•L−1 (versus the product label claim of 0.454 mmol/L for all brands). Recovery tests on spiked orange juice samples gave very good recovery (99-104%). The accuracy of the SWV results was validated by conducting parallel experiments using high-performance liquid chromatography (HPLC) on a C18 analytical column and absorbance (at 245 nm) detection. The mobile phase was 0.1% (v/v) formic acid in Milli-Q water, which was also used as the solvent for dilution of the samples. The measured AA content in the same commercial orange juice products ranged at 0.439-0.464 mmol•L−1. The HPLC method showed excellent linearity (R2 =0.998) and a detection limit (based on 3xsd) of 0.7 mmol L-1 AA. The reproducibility of replicate measurements was 2 % RSD. The SWV and HPLC results for the AA content of commercial orange juice samples were compared using F-test to compare precision and t-test to compare the means. The tests indicated that there was no significant difference (95% CL) in the precision and mean of the two datasets.