Usually, too many patients or users accessing a medical server can cause congestion and delay. To plan a system's capacity, we propose a queuing network model to estimate the performance index of a medical information system. First, we measure a set of the system response times with respect to the system specification. Second, by using the maximum likelihood function, we find the estimate of an equivalent service rate and the buffer size of a queue model from the simplified system. Third, with these queuing model parameters, we can simulate the performance of the medical information system. Then, before the implementation of the medical information system, we can estimate the performance of this system by means of a typical queuing simulation tool, such as QNAT.