<p>Facing various pieces of information disclosed by the system upon arrival, customers often exhibit different strategic responses. In this paper, customers’ strategic behavior is studied in a Markovian queue with Bernoulli-type working vacations. Upon completion of a service, the server starts a working vacation if the system is empty. If the system is found to be non-empty, the server takes a working vacation with a certain probability. During a working vacation, the server provides service at a lower service rate. Upon arrival, each customer decides whether to join the system or not based on the information disclosed and a reward-cost structure. The authors study the equilibrium balking strategies of customers at two information levels. For the fully observable case, the authors derive the two-dimensional threshold strategies, under which customers behave accordingly in the regular state and the working vacation state. For the partially observable case, the authors obtain a threshold strategy that completely depends on the queue length of the system. The influence of input parameters on the equilibrium strategies is discussed by numerical examples. Sensitivity analysis shows that reducing the vacation probability or rising the vacation rate will encourage more customers to join the system, thereby improving the system throughput. In addition, the disclosure of real-time server state information will also improve the system throughput.</p>