Statistical characteristics of discharge magnitude are significant for insulation diagnosis by monitoring partial discharge (PD) in power apparatuses. However, hitherto, investigations on the PD magnitude distribution in SF₆ gas under direct current voltage are insufficient. A current pulse measuring circuit and a ultra-high-frequency (UHF) antenna were simultaneously employed to investigate characteristics of positive corona in 0.3 MPa SF₆ gas. It is found that compared to the UHF antenna, the common measurement impedance used in laboratories caused several current pulses overlapped, resulting in pulse burst phenomenon. The pulse burst was easily mistaken as a single PD, which actually contained several PDs, thereby negatively affecting obtaining a real discharge magnitude distribution. Moreover, the discharge magnitude distributions were compared when a pulse burst was treated as a single PD and several PDs, respectively. It is found the former and latter one had forms of non-skewed two-parameter Weibull distribution with large-scale parameters and positively skewed two-parameter Weibull distribution with small-scale parameters, respectively. Based on this understanding, the change of the distributions with time was investigated, and the results showed that the distribution would have a permanent form of two-parameter Weibull distribution after a 24-h transition time.