The amortized analysis is a useful tool for analyzing the time-complexity of performing a sequence of operations. The disk scheduling problem involves a sequence of requests in general. In this paper, the performances of representative disk scheduling algorithms, SSTF, SCAN, and N-StepSCAN, are analyzed in the amortized sense. A lower bound of the amortized complexity for the disk scheduling problem is also derived. According to our analysis, SCAN is not only better than SSTF and N-StepSCAN, but also an optimal algorithm. Various authors have studied the disk scheduling problem based on some probability models and concluded that the most acceptable performance is obtained from SCAN. Our result therefore supports their conclusion.