Human mesenchymal stromal cell (hMSC) is a potential target for cell and gene therapy-based approaches against a variety of different diseases. Whilst cationic lipofection has been widely experimented, the Nucleofector technology is a relatively new non-viral transfection method designed for primary cells and hard-to-transfect cell lines. Herein, we compared the efficiency and viability of nucleofection with cationic lipofection, and used the more efficient transfection method, nucleofection, to deliver a construct of minimalistic, immunologically defined gene expression encoding the erythropoietin (MIDGE-EPO) into hMSC. MIDGE construct is relatively safer than the viral and plasmid expression systems as the detrimental eukaryotic and prokaryotic gene and sequences have been eliminated. Using a plasmid encoding the luciferase gene, we demonstrated a high transfection efficiency using the U-23 (21.79 ± 1.09%) and C-17 (5.62 ± 1.09%) pulsing program in nucleofection. The cell viabilities were (44.93 ± 10.10)% and (21.93 ± 5.72)%, respectively 24 h post-nucleofection. On the other hand, lipofection treatment only yielded less than 0.6% efficiencies despite showing higher viabilities. Nucleofection did not affect hMSC renewability, immunophenotype and differentiation potentials. Subsequently, we nucleofected MIDGE-EPO using the U-23 pulsing program into hMSC. The results showed that, despite a low nucleofection efficiency with this construct, the EPO protein was stably expressed in the nucleofected cells up to 55 days when determined by ELISA or immunocytochemical staining. In conclusion, nucleofection is an efficient non-viral transfection approach for hMSC, which when used in conjunction with a MIDGE construct, could result in extended and stable transgene expression in hMSC.