Particle injections occur frequently inside 10 Re during geomagnetic storms. They are commonly associated with bursty bulk flows or plasma sheet bubbles transported from the tail to the inner magnetosphere. Although observations and theoretical arguments have suggested that they may have an important role in storm time dynamics, this assertion has not been addressed quantitatively. In this paper, we investigate which process is dominant for the storm time ring current buildup: large-scale enhanced convection or localized bubble injections. We use the Rice Convection Model-Equilibrium (RCM-E) to model a series of idealized storm main phases. The boundary conditions at 14–15 Re on the nightside are adjusted to randomly inject bubbles to a degree roughly consistent with observed statistical properties. A test particle tracing technique is then used to identify the source of the ring current plasma. We find that the contribution of plasma sheet bubbles to the ring current energy increases from ~20% for weak storms to ~50% for moderate storms and levels off at ~61% for intense storms, while the contribution of trapped particles decreases from ~60% for weak storms to ~30% for moderate and ~21% for intense storms. The contribution of nonbubble plasma sheet flux tubes remains ~20% on average regardless of the storm intensity. Consistent with previous RCM and RCM-E simulations, our results show that the mechanisms for plasma sheet bubbles enhancing the ring current energy are (1) the deep penetration of bubbles and (2) the bulk plasma pushed ahead of bubbles. Both the bubbles and the plasma pushed ahead typically contain larger distribution functions than those in the inner magnetosphere at quiet times. An integrated effect of those individual bubble injections is the gradual enhancement of the storm time ring current. We also make two predictions testable against observations. First, fluctuations over a time scale of 5–20 min in the plasma distributions and electric field can be seen in the central ring current region for the storm main phase. We find that the plasma pressure and the electric field EY there vary over about 10%–30% and 50%–300% of the background values, respectively. Second, the maximum plasma pressure and magnetic field depression in the central ring current region during the main phase are well correlated with the Dst index.