PLX3397 is an orally administrable CSF1 receptor inhibitor that is expected to be a therapeutic agent for cell tumors such as tendon synovial giant cell tumors. Administration of high doses of PLX3397 is known to eliminate microglia selectively and may affect neuronal functions via the loss of microglia. The elimination of microglia has been known to affect exploratory tasks involving aversive stimuli. However, the effect of the microglial elimination on the ability of reward-based reinforcement learning has been unknown. In this study, C57BL6 mice were administrated with PLX3397 for three weeks and were tested with a 5-armed bandit task (5-ABT). 5-ABT is an exploratory operant conditioning task in which each of the five choices has a different reward probability. The task contains three subtasks: one in which all choices had a reward probability of 30% (ALL), a subtask in which only one choice had a reward probability of 50% and the rest had a reward probability of 0% (BIT), and a subtask in which the correct choice in BIT was reversed and four options had a reward probability of 30% (REV). As a result, there was no significant difference in the entropy of choice (exploration pattern) in the subtask ALL between the PLX3397-treated group and the control group. There was no significant difference between the groups in the number of steps required for the BIT and REV subtasks to exceed 50% correct. We also estimated the learning rates of the mice by fitting the behavioral sequences to a Q-learning model and found no significant differences between the groups. These results suggest that microglial elimination has no significant impact on reinforcement learning ability.