Abstract: A recently released XMM-Newton note revealed a significant calibration issue between nuclear spectroscopic telescope array (NuSTAR) and XMM-Newton European Photon Imaging Camera (EPIC) and provided an empirical correction to the EPIC effective area. To quantify the bias caused by the calibration issue in the joint analysis of XMM-NuSTAR spectra and verify the effectiveness of the correction, in this work, we perform joint-fitting of the NuSTAR and EPIC-pn spectra for a large sample of 104 observation pairs of 44 X-ray bright active galactic nuclei (AGN). The spectra were extracted after requiring perfect simultaneity between the XMM-Newton and NuSTAR exposures (merging good time intervals (GTIs) from two missions) to avoid bias due to the rapid spectral variability of the AGN. Before the correction, the EPIC-pn spectra are systematically harder than the corresponding NuSTAR spectra by \Delta \varGamma \sim 0.1 , subsequently yielding significantly underestimated cutoff energy E_cut and the strength of reflection component R when performing joint-fitting. We confirm that the correction is highly effective and can commendably erase the discrepancy in best-fit Γ, E_cut, and R. We thus urge the community to apply the correction when joint-fitting XMM-NuSTAR spectra, but note that the correction is limited to 3–12 keV and therefore not applicable when the soft X-ray band data are included. Besides, we show that as merging GTIs from two missions would cause severe loss of NuSTAR net exposure time, in many cases, joint-fitting yields no advantage compared with utilizing NuSTAR data alone. Finally, We present a technical note on filtering periods of high background flares for XMM-Newton EPIC-pn exposures in the small window (SW) mode.