We consider a multi- server queueing inventory system with two types of customers: Type I and Type II. Type II customers are virtual ones. Arrival of both Type I and Type II customers follow two independent Poisson processes. Type I are to be served by one of the servers and service time is assumed to be exponentially distributed. Type II customer may be served by a Type I customer having already been served and ready to act as a server or by one of the servers with exponentially distributed service time. Type I customer has non preemptive priority over Type II. Type II is served by a Type I only if inventory is available after attaching inventory to the existing Type-I customers available in the system. Type II is served by a Type I with probability p on completion of latter’s service and with complementary probability q=1−p served Type I leaves the system without serving a Type-II. Fresh arrivals of both type of customers are permitted to join the system only when excess inventory, which is defined as the difference between on hand inventory and number of busy servers is positive. System capacity for Type I is limited, where as Type II has unlimited waiting area. When inventory level drops to c+s , an order for replenishment is placed to bring the level to c+S . The ordered items are received after a random amount of time which is exponentially distributed. A revenue function is constructed and effect of probability p on the revenue function for single server, two and three servers is numerically analyzed. Effects of parameter β of the exponentially distributed lead time, on the loss rate of customers and revenue function are numerically analyzed.