Abstract We present results—cool-down, energization, and persistent-mode operation—of a solid-nitrogen (SN2)-cooled, magnesium diboride (MgB₂) small-scale test coil. The test coil, immersed in a volume of solid nitrogen at 6 K, successfully operated in persistent-mode at 108 A for a period of 5 d. Although designated a ‘persistent-mode’ coil, its center field was measured to decay at a rate of <0.6 ppm h⁻¹, which is still considered low enough to meet the temporal stability requirement of <0.1 ppm h⁻¹, for most magnetic resonance imaging magnets. This decay rate translates to a calculated circuit resistance of <1.79 × 10^–12 Ω, which is mainly from one MgB₂-MgB₂ joint in the circuit. However, when the coil temperature increased from 6 to 16 K, the field had dropped by 0.33%: we believe this was caused by the change of magnetization in the MgB₂ superconductor, which in turn decreased a screening-current field (SCF) at the magnet center. We performed a finite element analysis with a simplified numerical model based on H formulation to verify whether magnetization-induced SCF is responsible for this 0.33% drop. Indeed, the model shows that the change of magnetization, i.e. screening-current reduction and current density redistribution, happens during temperature-cycle-induced J_c (T) variation, and thus affects the center magnetic field. However, the J_c (T) variation in the 2nd cycle had little effect on MgB₂ magnetization and thus had negligible magnetic field change.