Copper matrix-matched calibration samples were prepared by doping high purity copper powders (99.999%, 500–600 μm) with multi-element standard solutions with a total of 72 elements, and then the relative sensitivity factors (RSFs) of the 72 elements in both microsecond-pulsed and continuous direct current (dc) modes were determined by using Element GD type glow discharge mass spectrometry (GD-MS). The ability of pulsed and continuous dc-GD-MS to quantitatively determine trace elements in copper was demonstrated after calibration using the RSFs. The results showed that the uncertainty level of GD-MS analysis after calibration was comparable with that of ICP-MS and ICP-OES analysis. In order to standardize calibration sample preparation procedures, the doping concentration, pressed powder tablet compactness and tablet homogeneity were optimized. Good recovery (85% to 115%) was obtained. Good pressed powder tablet compactness and tablet homogeneity were obtained, which were expressed as relative standard deviations (<5%) of 6 times GD-MS analysis. Satisfactory repeatability of about 15% (RSD) for continuous and 10% (RSD) for pulsed dc-GD-MS was obtained by strictly controlling the intensity of the matrix to keep it stable when replacing graphite parts. Good reproducibility (RSD < 10%) was obtained after timely detector calibration, so the determined RSFs could remain constant and valid in an instrument for a long time. Under the optimized conditions mentioned above, the RSFs of the total 72 elements in both continuous and pulsed modes were determined and compared. The comparison of RSFs showed that the general trends of the two sets of RSFs were very similar; however, the RSFs in pulsed mode were 2–40% lower than those in continuous mode in most cases. Therefore, a dedicated RSF table for each mode is recommended to achieve accurate and quantitative analysis. The comparison of RSFs in this work with standard RSFs provided by the manufacturer of Element GD revealed that the accepted uncertainty of a factor of two was met for almost all investigated elements in both modes. The accuracy of the RSFs was validated by analyzing three Cu-matrix CRMs. The relative expanded uncertainties (U_rel, k = 2) of measured results by using pulsed dc-GD-MS, including the uncertainty of measurement repeatability and RSF, were less than 20%, which was much lower than those obtained by continuous dc-GD-MS (U_rel < 30%).